{"variant":"public","generated":"2026-10-12T17:39:20.785Z","fields":[{"slug":"document-image-analysis","name":{"ja":"文書画像解析","en":"Document Image Analysis"}},{"slug":"information-retrieval","name":{"ja":"情報検索","en":"Information Retrieval"}},{"slug":"pattern-recognition","name":{"ja":"パターン認識","en":"Pattern recognition"}},{"slug":"object-recognition","name":{"ja":"物体認識","en":"Object Recognition"}},{"slug":"nearest-neighbor-search","name":{"ja":"最近傍探索","en":"Nearest Neighbor Search"}},{"slug":"computational-photography","name":{"ja":"コンピュテーショナル フォトグラフィ","en":"Computational Photography"}},{"slug":"reading-life-log","name":{"ja":"リーディングライフログ","en":"Reading-Life Log"}},{"slug":"3d-reconstruction","name":{"ja":"三次元再構築","en":"3D Reconstruction"}},{"slug":"visually-impaired-assistance","name":{"ja":"視覚障害者支援","en":"Visually impaired assistance"}},{"slug":"plant-image-processing","name":{"ja":"植物画像処理","en":"Plant image processing"}},{"slug":"deep-learning","name":{"ja":"ディープラーニング","en":"Deep Learning"}},{"slug":"learning-analytics","name":{"ja":"学習解析","en":"Learning Analytics"}},{"slug":"sensing","name":{"ja":"センシング","en":"Sensing"}},{"slug":"quantum-computing","name":{"ja":"量子コンピューティング","en":"Quantum Computing"}},{"slug":"digital-watermarking","name":{"ja":"電子透かし","en":"Digital Watermarking"}},{"slug":"comic-computing","name":{"ja":"コミック工学","en":"Comic Computing"}},{"slug":"document-image-retrieval","name":{"ja":"文書画像検索","en":"Document Image Retrieval"}},{"slug":"crowdsourcing","name":{"ja":"クラウドソーシング","en":"Crowdsourcing"}},{"slug":"activity-recognition","name":{"ja":"行動認識","en":"Activity Recognition"}},{"slug":"sign-language-recognition","name":{"ja":"手話認識","en":"Sign Language Recognition"}},{"slug":"wellbeing","name":{"ja":"ウエルビーイング","en":"Wellbeing"}},{"slug":"medical-ai","name":{"ja":"医療AI","en":"Medical AI"}},{"slug":"chocolate","name":{"ja":"チョコレート","en":"Chocolate"}},{"slug":"handwriting","name":{"ja":"手書き","en":"Handwriting"}},{"slug":"satellite-image-analysis","name":{"ja":"衛星画像解析","en":"Satellite Image Analysis"}}],"publications":[{"key":"Stricker2026_MDPI","id":"2125","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Marco Stricker","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"CBEN - A Multimodal Machine Learning Dataset for Cloud-Robust Remote Sensing Image Understanding"},"journal":{"ja":"MDPI Electronics"},"book_title":"Electronics","volume":"15","pages":"2927","year":2026,"month":7,"publisher":"MDPI","review":"reviewed","fields":["satellite-image-analysis","deep-learning"],"doi":"10.3390/electronics15132927","url":"https://www.mdpi.com/2079-9292/15/13/2927","abstract":{"ja":"Clouds frequently degrade optical satellite imagery, limiting the reliability of remote sensing models. However, in the literature, cloud-free analyses are often performed by excluding cloudy images from machine learning datasets and methods. This restricts their usefulness in time-critical scenarios such as disaster response, where waiting for cloud-free imagery is impractical. Cloud removal can mitigate this issue, but methods remain imperfect and may introduce visual artifacts. Therefore, it is desirable to develop cloud-robust methods by combining optical imagery with radar data, a modality unaffected by clouds. While datasets for machine learning combine optical and radar data, most researchers exclude cloudy images from training and evaluation. We identify this exclusion as a limitation that reduces applicability to cloudy scenarios and address it by introducing CloudyBigEarthNet (CBEN), a dataset of paired optical and radar images containing cloud occlusions for land-use and land-cover classification. Using average precision (AP), we show that state-of-the-art methods trained on clear-sky optical and radar data suffer performance drops of between 23.8 and 33.4 AP points when tested on cloudy imagery. We adapt these methods using cloudy images during training and improve AP on cloudy test cases by 17.2 to 28.7 AP points. Code and dataset have been published."}},{"key":"Dengel2026","id":"2121","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Andreas Dengel","en":""},{"ja":"Laurence Devillers","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Andrew Vargo (Eds)","en":""}],"title":{},"book_title":"The Future of Education with AI","year":2026,"month":1,"publisher":"Springer Nature","review":"not_reviewed","fields":["sensing","learning-analytics"]},{"key":"Minatani2026_ICGG","id":"2124","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Kazunori Minatani","en":""},{"ja":"Kosei Takaishi","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Photo-based 2.5D Reliefs for TeachingPerspective and Depth to the Visually Impaired"},"journal":{"ja":"Proc. the 22nd International Conference on Geometry and Graphics (ICGG 2026)"},"year":2026,"month":8,"location":"Zagreb, Croatia","review":"reviewed","presen_type":"oral","fields":["visually-impaired-assistance"],"abstract":{"ja":"The visually impaired have limited access to understanding\nvision-based spatial phenomena such as linear perspective, depth cues,\nand occlusion, which sighted people routinely use to interpret images.\nWhile prior work has highlighted the importance of making perspec-\ntival images accessible to blind learners, it has provided few concrete,\ngeneralizable strategies for systematically teaching pictorial space. We\npropose a practical and scalable approach that uses photographs (rather\nthan canonical paintings) as the source of tactile educational material.\nOur pipeline estimates depth from a photograph to generate a 2.5D\nrelief height ?eld, then fabricates tactile models via 3D printing. This\nphoto-based route is (i) general, because photographs can be captured\non-demand to exemplify speci?c spatial cues in a staged and progres-\nsively structured manner, and (ii) practical, because it reduces manual\nproduction burden compared to handcrafted tactile translations. We in-\ntegrate lessons from tactile picture perception research and present a\nstepwise pedagogy that begins with monocular perspective cues and ex-\ntends toward binocular depth concepts. We further propose a qualitative\nevaluation framework based on think-aloud protocols, semi-structured\ninterviews, and thematic analysis to assess educational e?cacy of tactile\nphotographs."}},{"key":"Minatani2026_DSAI","id":"2123","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Kazunori Minatani","en":""},{"ja":"Tetsuya Watanabe","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Eriko Shima","en":""}],"title":{"ja":"Retention in a Series of Online Symposia with Tangible 3D Models for the Visually Impaired: Survival and Cohort Analyses for Accessible Edutainment Outreach"},"journal":{"ja":"Proc. the 12th International Conference on Software Development and Technologies for Enhancing Accessibility and Fighting Info-exclusion (DSAI 2026)"},"year":2026,"month":6,"location":"Alexandroupolis, Greece","review":"reviewed","presen_type":"oral","fields":["visually-impaired-assistance"],"awards":{"en":["Best Paper Award Winner"]},"abstract":{"ja":"This paper examines retention in an episodic, semiannual online\nsymposium series on 3D learning for the SDGs that combines\nZoom-based delivery with tangible learning aids for accessible outreach.\nUsing attendance records from 10 online symposia (S2?S11;\n? = 784), retention is analyzed with Kaplan?Meier survival estimation,\nduration distributions, cohort retention, and a comparison\nbetween blind and low-vision (BLV) and non-disabled registrants.\nOverall retention shows a steep early drop: 526 participants (67.1%)\nattended exactly one symposium, while 198 (25.3%) attended the\nfinal observed symposium and are right-censored. At the same\ntime, re-engagement is substantial: 135 of 258 repeat attendees\n(52.3%) returned after skipping at least one symposium. The groupdisaggregated\nanalysis shows that BLV registrants were markedly\nmore persistent than non-disabled registrants, with higher survival\nacross attended events, higher next-session retention (40.6%\nvs. 16.5%), and a larger repeat-attendance share (54.8% vs. 24.7%).\nAlthough MOOC completion rates are only a reference and not\na direct analogue, the BLV retention profile is notably stronger\nthan the high-attrition baselines commonly reported for online\nlearning. These findings suggest that webinar-based participation,\nwhen paired with tactile access pathways such as mailed 3D models,\ncan sustain participation particularly well for BLV learners while\nstill leaving important tactile, logistical, and first-session design\nchallenges unresolved."}},{"key":"Minatani2026","id":"2100","type":"intl_conf","language":"JA","show":"show","authors":[{"ja":"Kazunori Minatani","en":""},{"ja":"Kosei Takaishi","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Photo-Based 2.5D Reliefs for Teaching Perspective and Depth to the Visually Impaired"},"journal":{"ja":"Proceedings of the 22nd International Conference on Geometry and Graphics"},"year":2026,"month":4,"location":"Zagreb, Croatia","review":"reviewed","fields":[]},{"key":"Harimakoto2026_MIRU","id":"2132","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"梁誠","en":""},{"ja":"内海ゆづ子","en":""},{"ja":"岩村雅一","en":""}],"title":{"ja":"3次元点群間の差異説明のための比較アーキテクチャ"},"journal":{"ja":"第29回 画像の認識・理解シンポジウム（MIRU2026）予稿集"},"presen_id":"IS2-155","num_pages":"5","year":2026,"month":8,"location":"出島メッセ長崎","review":"not_reviewed","presen_type":"poster","fields":[]},{"key":"Imayoshi2026","id":"2127","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"今吉 嶺太","en":"Reita Imayoshi"},{"ja":"内海 ゆづ子","en":"Yuzuko Utsumi"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"3D Gaussianの時間的対応関係に基づく植物器官の時系列追跡","en":"Time-Series Tracking of Plant Organs Based on the Temporal Correlation of 3D Gaussian"},"journal":{"ja":"第29回 画像の認識・理解シンポジウム（MIRU2026）予稿集"},"presen_id":"IS3-038","pages":"97","num_pages":"5","year":2026,"month":8,"location":"出島メッセ長崎","review":"not_reviewed","presen_type":"poster","fields":["3d-reconstruction","plant-image-processing"],"abstract":{"ja":"本研究では，収量予測や生育診断において重要となる，植物の3 次元時系列データにおける個々の器官の追跡を目的とする．植物は非剛体で成長にトポロジー変化を伴うため従来の物体追跡手法を適用できず，器官追跡は実現されていない．そこで本研究では，3 次元時系列植物データを3D Gaussian で表現し，器官を表すGaussian 群の時系列の変化によって，追跡を行う手法を提案する．","en":"This study aims to track individual plant organs in 3D time-series data, which is crucial for yield prediction and growth diagnosis. Because plants are non-rigid objects whose growth involves topological changes, conventional object tracking methods cannot be applied, and organ tracking has not yet been achieved. Therefore, in this study, we represent 3D time-series plant data as 3D Gaussians and propose a method for tracking based on the temporal changes in the group of Gaussians representing the organs."}},{"key":"Kawagoe2026","id":"2128","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"川越壮","en":""},{"ja":"内海ゆづ子","en":""},{"ja":"岩村雅一","en":""}],"title":{"ja":"クラス名に依存しない視覚的記述子探索によるCLIPのOOP適応"},"journal":{"ja":"第29回 画像の認識・理解シンポジウム（MIRU2026）予稿集"},"presen_id":"IS2-163","pages":"82","year":2026,"month":8,"location":"出島メッセ長崎","review":"not_reviewed","presen_type":"poster","fields":["deep-learning","object-recognition"]},{"key":"Sugaya2026","id":"2129","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"菅谷 友稀","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"三輪 由佳","en":""},{"ja":"岩村 雅一","en":""}],"title":{"ja":"ブドウ果粒の三次元位置情報に基づく摘粒対象提示アプリケーションの開発"},"journal":{"ja":"第29回 画像の認識・理解シンポジウム（MIRU2026）予稿集"},"presen_id":"DS-08","num_pages":"111","year":2026,"month":8,"review":"not_reviewed","presen_type":"poster","fields":["3d-reconstruction","plant-image-processing"],"abstract":{"ja":"本研究では, ブドウ摘粒作業の習得支援を目的として, 動画撮影から摘粒対象の提示までを自動的に完結させる摘粒支援アプリケーションを開発した. 三次元点群処理による三次元再構成と球体検出によって各果粒の三次元位置を推定し, 摘粒すべき果粒の選択を組合せ最適化問題として定式化して遺伝的アルゴリズムを適用することで摘粒対象を決定する. 決定した摘粒対象をインタラクティブな3D モデル上に可視化することで, ユーザが直感的に摘粒箇所を確認できる. 評価実験では, アプリを使用することで初心者でも熟練者と同程度の果粒間距離を実現できることを確認した."}},{"key":"Yagihara2026","id":"2133","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"八木原直樹","en":"Naoki Yagihara"},{"ja":"内海ゆづ子","en":"Yuzuko Utsumi"},{"ja":"山内大輔","en":"Daisuke Yamauchi"},{"ja":"峰雪芳宣","en":"Yoshinobu Mineyuki"},{"ja":"唐原一郎","en":"Ichirou Karahara"},{"ja":"星野真人","en":"Masato Hoshino"},{"ja":"上杉健太朗","en":"Kentaro Uesugi"},{"ja":"鈴木芳生","en":"Yoshio Suzuki"},{"ja":"竹内晃久","en":"Akihisa Takeuchi"},{"ja":"岩村雅一","en":"Masakazu Iwamura"}],"title":{"ja":"無染色X線μCTにおける少数注釈での密充填植物細胞のインスタンス分割","en":"Instance segmentation of densely packed plant cells in stain-free X-ray μCT under few annotations"},"journal":{"ja":"第29回 画像の認識・理解シンポジウム（MIRU2026）予稿集"},"presen_id":"IS2-253","num_pages":"5","year":2026,"month":8,"location":"出島メッセ長崎","review":"not_reviewed","presen_type":"poster","fields":["plant-image-processing"],"abstract":{"ja":"無染色X線マイクロCTは植物組織を非破壊・細胞単位で撮像できるが、細胞壁が低コントラストで密充填のため、少ない注釈で各細胞を分割するのは難しい。本研究はこの少数注釈かつ境界不明瞭な設定に取り組む。細胞壁に依存しないflow系モデルに部分ラベル・一貫性正則化・合成教師データを組み合わせ、少数注釈での分割に対応する。さらに、μCT画像に細胞壁の情報がそもそも乏しく学習が難しいことを定量的に示す。"}},{"key":"Sakai2026_CVIM","id":"2122","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"酒井 柊斗","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"岩村 雅一","en":""},{"ja":"米倉 崇晃","en":""},{"ja":"中井 朋則","en":""},{"ja":"山内 大輔","en":""},{"ja":"唐原 一郎","en":""},{"ja":"峰雪 芳宣","en":""},{"ja":"星野 真人","en":""},{"ja":"上杉 健太朗","en":""},{"ja":"小塚 俊明","en":""}],"title":{"ja":"生育段階の異なるキクタニギクの3次元小花配列の推定"},"journal":{"ja":"情報処理学会 研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2026-CVIM-246","presen_id":"40","num_pages":"8","year":2026,"month":5,"publisher":"情報処理学会","location":"富山県民会館","review":"not_reviewed","presen_type":"poster","fields":[],"url":"https://cvim.ipsj.or.jp/index.php?id=246p"},{"key":"Harimakoto2026_CVIM","id":"2131","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"梁誠","en":""},{"ja":"内海ゆづ子","en":""},{"ja":"岩村雅一","en":""}],"title":{"ja":"3次元点群データ間の差異に対する自然言語記述の生成"},"journal":{"ja":"情報処理学会研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2026-CVIM-246","number":"39","pages":"1-7","year":2026,"month":5,"location":"富山県民会館","review":"not_reviewed","presen_type":"poster","fields":[]},{"key":"Ida2026_GC","id":"2103","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"井田圭哉","en":""},{"ja":"杉本美","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"官能評価値の予測における表形式基盤モデルの有用性の調査"},"journal":{"ja":"2026年電子情報通信学会総合大会講演論文集"},"presen_id":"D-12-29","num_pages":"1","year":2026,"month":3,"review":"not_reviewed","fields":["chocolate"]},{"key":"Ryuu2026","id":"2104","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"劉 少楠","en":""},{"ja":"岩下雄一郎","en":""},{"ja":"仲子聡一郎","en":""},{"ja":"岩村雅一","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"連続時間型の拡散モデルを用いた混合型時系列 EHR の生成"},"journal":{"ja":"2026年電子情報通信学会総合大会講演論文集"},"presen_id":"D-19-16","year":2026,"month":3,"review":"not_reviewed","presen_type":"oral","fields":["medical-ai"]},{"key":"Takahashi2026","id":"2101","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"高橋歩夢","en":""},{"ja":"井田圭哉","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"官能評価値予測における表形式基盤モデルのファインチューニング有効性についての調査"},"journal":{"ja":"2026年電子情報通信学会総合大会講演論文集"},"presen_id":"D-12-28","num_pages":"1","year":2026,"month":3,"review":"not_reviewed","presen_type":"oral","fields":["chocolate"]},{"key":"Saimyou2026_GC","id":"2084","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"最明 新太","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"三輪 由佳","en":""},{"ja":"岩村 雅一","en":""},{"ja":"大山 克己","en":""}],"title":{"ja":"異常検知を用いたブドウの葉の病変検出"},"journal":{"ja":"電子情報通信学会総合大会"},"presen_id":"D-12-12","year":2026,"month":3,"location":"九州産業大学","review":"not_reviewed","presen_type":"oral","fields":["pattern-recognition"]},{"key":"Kosei2026","id":"2097","type":"bachelors_thesis","language":"JA","show":"show","authors":[{"ja":"高石 康世","en":"Takaishi Kosei"}],"title":{"ja":"深度を利用した2.5D レリーフによる視覚障害者のための触覚写真の表現力拡張"},"year":2026,"month":3,"review":"not_reviewed","fields":[]},{"key":"Iste2026","id":"2105","type":"other","language":"EN","show":"show","authors":[{"ja":"","en":"Adrian Iste"},{"ja":"","en":"Kazuki Nishizawa"},{"ja":"","en":"Chisa Tanaka"},{"ja":"","en":"Andrew Vargo"},{"ja":"","en":"Anna Scius-Bertrand"},{"ja":"","en":"Andreas Fischer"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Prediction of Grade, Gender, and Academic Performance of Children and Teenagers from Handwriting Using the Sigma-Lognormal Model"},"journal":{"en":"arXiv"},"num_pages":"18","year":2026,"month":3,"review":"not_reviewed","fields":["handwriting"],"url":"arXiv:2603.11519"},{"key":"Iwashita2026","id":"2108","type":"other","language":"EN","show":"show","authors":[{"ja":"","en":"Yuichiro Iwashita"},{"ja":"","en":"Ahtisham Fazeel Abbasi"},{"ja":"","en":"Koichi Kise"},{"ja":"","en":"Andreas Dengel"},{"ja":"","en":"Muhammad Nabeel Asim"}],"title":{"en":"A Large-Scale Comparative Analysis of Imputation Methods for Single-Cell RNA Sequencing Data"},"journal":{"en":"arXiv"},"year":2026,"month":3,"review":"not_reviewed","fields":["medical-ai"],"doi":"10.48550/arXiv.2603.24626","url":"https://doi.org/10.48550/arXiv.2603.24626"},{"key":"Liu2026","id":"2107","type":"other","language":"EN","show":"show","authors":[{"ja":"","en":"Shaonan Liu"},{"ja":"","en":"Yuichiro Iwashita"},{"ja":"","en":"Soichiro Nakako"},{"ja":"","en":"Masakazu Iwamura"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"CDMT-EHR: A Continuous-Time Diffusion Framework for Generating Mixed-Type Time-Series Electronic Health Records"},"journal":{"en":"arXiv"},"num_pages":"8","year":2026,"month":3,"review":"not_reviewed","fields":["medical-ai"],"doi":"10.48550/arXiv.2603.23719","url":"https://doi.org/10.48550/arXiv.2603.23719"},{"key":"Tanaka2026","id":"2106","type":"other","language":"EN","show":"show","authors":[{"ja":"","en":"Chisa Tanaka"},{"ja":"","en":"Andrew Vargo"},{"ja":"","en":"Anna Scius-Bertrand"},{"ja":"","en":"Andreas Fischer"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"From Pen Strokes to Sleep States: Detecting Low-Recovery Days Using Sigma-Lognormal Handwriting Features"},"journal":{"en":"arXiv"},"num_pages":"16","year":2026,"month":3,"review":"not_reviewed","fields":["handwriting"],"url":"https://arxiv.org/abs/2603.11512"},{"key":"Buoy2026","id":"2111","type":"other","language":"EN","show":"show","authors":[{"ja":"","en":"Rina Buoy"},{"ja":"","en":"Dylan berkamp Fouepe Dongmo"},{"ja":"","en":"Vesal Khean"},{"ja":"","en":"Simone Marinai"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Towards Non-Latin Text and Layout Personalization for Enhanced Readability"},"journal":{"en":"arXiv"},"year":2026,"month":2,"review":"not_reviewed","fields":["document-image-analysis"],"url":"https://doi.org/10.48550/arXiv.2603.00688"},{"key":"Kong2026_arXiv_Khmer","id":"2109","type":"other","language":"EN","show":"show","authors":[{"ja":"","en":"Marry Kong"},{"ja":"","en":"Rina Buoy"},{"ja":"","en":"Sovisal Chenda"},{"ja":"","en":"Nguonly Taing"},{"ja":"","en":"Masakazu Iwamura"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Towards Khmer Scene Document Layout Detection"},"journal":{"en":"arXiv"},"num_pages":"17","year":2026,"month":2,"review":"not_reviewed","fields":["document-image-analysis"],"url":"https://doi.org/10.48550/arXiv.2603.00707"},{"key":"Kong2026_arXiv_Universal","id":"2110","type":"other","language":"EN","show":"show","authors":[{"ja":"","en":"Marry Kong"},{"ja":"","en":"Rina Buoy"},{"ja":"","en":"Sovisal Chenda"},{"ja":"","en":"Nguonly Taing"},{"ja":"","en":"Masakazu Iwamura"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Towards Universal Khmer Text Recognition"},"journal":{"en":"arXiv"},"num_pages":"17","year":2026,"month":2,"review":"not_reviewed","fields":["document-image-analysis"],"url":"https://doi.org/10.48550/arXiv.2603.00702"},{"key":"Stricker2026_arXiv","id":"2112","type":"other","language":"EN","show":"show","authors":[{"ja":"","en":"Marco Stricker"},{"ja":"","en":"Masakazu Iwamura"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"CBEN -- A Multimodal Machine Learning Dataset for Cloud Robust Remote Sensing Image Understanding"},"journal":{"en":"arXiv"},"year":2026,"month":2,"review":"not_reviewed","fields":["satellite-image-analysis"],"url":"https://doi.org/10.48550/arXiv.2602.12652"},{"key":"Siddhad2025","id":"2064","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Gourav Siddhad","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Partha Pratim Roy","en":""}],"title":{"ja":"Enhancing EEG Signal-Based Emotion Recognition with Synthetic Data: Diffusion Model Approach"},"journal":{"ja":"IEEE Transactions on Artificial Intelligence"},"year":2025,"month":12,"publisher":"IEEE","review":"reviewed","fields":["sensing","deep-learning"],"doi":"10.1109/TAI.2025.3641576","abstract":{"ja":"Emotions are crucial in human life, influencing perceptions, relationships, behaviour, and choices. Emotion recognition using Electroencephalography (EEG) in the Brain-Computer Interface (BCI) domain presents significant challenges, particularly the need for extensive datasets. This study aims to generate synthetic EEG samples similar to real samples but distinct by augmenting noise to a conditional denoising diffusion probabilistic model, thus addressing the prevalent issue of data scarcity in EEG research. The proposed method is tested on the DEAP and SADT datasets, showcasing up to 5.6\\% improvement in classification accuracy when using synthetic data with DEAP and similar positive results with SADT. This is higher compared to the traditional Generative Adversarial Network (GAN) based and Denoising Diffusion Probabilistic Model (DDPM) based approaches. This study further evaluates the effectiveness of state-of-the-art classifiers on EEG data, employing both real and synthetic data with varying noise levels, and utilizes t-SNE and SHAP for detailed analysis and interpretability. The proposed diffusion-based approach for EEG data generation appears promising in refining the accuracy of emotion recognition systems and marks a notable contribution to EEG-based emotion recognition."}},{"key":"Singh2025","id":"2047","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Shrey Singh","en":""},{"ja":"Prateek Keserwani","en":""},{"ja":"Katsufumi Inoue","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Partha Pratim Roy","en":""}],"title":{"ja":"Revisiting I3D for Sign Language Recognition"},"journal":{"ja":"IEICE Transactions on Information and Systems"},"volume":"E108-D","number":"11","pages":"1359-1372","num_pages":"14","year":2025,"month":11,"publisher":"IEICE","review":"reviewed","fields":["sign-language-recognition"],"doi":"10.1587/transinf.2024EDP7297"},{"key":"Buoy2025_IJDAR","id":"2058","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"","en":"Rina Buoy"},{"ja":"","en":"Sovisal Chenda"},{"ja":"","en":"Nguonly Taing"},{"ja":"","en":"Marry Kong"},{"ja":"","en":"Masakazu Iwamura"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Addressing the Attention Drift Problem for Khmer Long Textline Recognition"},"journal":{"en":"International Journal on Document Analysis and Recognition (IJDAR)"},"year":2025,"month":9,"publisher":"Springer","address":"Berlin, Germany","review":"reviewed","fields":["document-image-analysis","pattern-recognition","deep-learning"],"doi":"10.1007/s10032-025-00554-6","abstract":{"ja":"An autoregressive (AR) decoder generates one character or token at a time, based on previous outputs (i.e., context). This approach inherently includes an internal language model, which can lead to higher recognition accuracy compared to a non-autoregressive (NAR) decoder. However, this internal language model can be a double-edged sword. For long textline inputs, such as Khmer textline images lacking word boundaries, the autoregressive (AR) decoding process can accumulate errors at each step. Once a critical error threshold is reached, the decoder may struggle to align the corresponding visual features with the decoded outputs. This misalignment, known as the attention drift problem, arises because many AR decoders rely on attention mechanisms for alignment. In this paper, we propose a robust Khmer textline recognition method based on a partially autoregressive (PAR) decoder. Our method effectively addresses the attention drift problem when dealing with long textline inputs and thus maintains high recognition accuracy. The experimental results demonstrate that the proposed method significantly improves recognition accuracy for Khmer long textlines by mitigating the attention drift issue. Compared to existing state-of-the-art (SoTA) methods, our approach establishes new SoTA performance. Further experiments on the Latin handwritten datasets, IAM and RIMES, validate the robustness of the proposed method."}},{"key":"Kise2025","id":"2052","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"黄瀬浩一","en":""}],"title":{"ja":"センシングとアクチュエーション技術による読みと学びの増強"},"journal":{"ja":"AI・データサイエンス論文集"},"volume":"6","number":"1","pages":"14-25","year":2025,"month":5,"review":"not_reviewed","fields":["reading-life-log","sensing","learning-analytics"],"doi":"10.11532/jsceiii.6.1_14"},{"key":"Iwatsuru2025","id":"2025","type":"journal_paper","language":"EN","show":"show_except_datafile","authors":[{"ja":"Kyoya Iwatsuru","en":"Kyoya Iwatsuru"},{"ja":"Shoya Ishimaru","en":"Shoya Ishimaru"},{"ja":"Andrew Vargo","en":"Andrew Vargo"},{"ja":"Koichi Kise","en":"Koichi Kise"}],"title":{"ja":"Enhancing Text Comprehension forLanguage Learning by GeneratingQuestions with Eye-Tracking and LLM","en":"Enhancing Text Comprehension forLanguage Learning by GeneratingQuestions with Eye-Tracking and LLM"},"journal":{"ja":"Activity and Behavior Computing","en":"Activity and Behavior Computing"},"year":2025,"month":4,"location":"Abu Dhabi","review":"reviewed","presen_type":"oral","fields":["reading-life-log","sensing"],"abstract":{"ja":"This paper presents a system that supports a reader by identifying\ndifficult paragraphs using eye-tracking implicitly and generating compre-\nhension questions with a Large Language Model (LLM). Our aim is to\nevaluate the potential of generative AI as a tool to augment human cog-\ngnitive capabilities. If readers answer the questions correctly, this confirms\nthat they understand the text. Conversely, if they answer incorrectly, it\nindicates a lack of understanding, allowing the system to provide explana-\ntions for the misunderstood parts, thereby improving the readers’ under-\nstanding of the text. Our experiment showed that half of the participants\nimproved their comprehension of the text. Further analysis revealed that\nthe motivation to read/use system is one of the most important factor to\nincrease the effectiveness of the proposed system.","en":"This paper presents a system that supports a reader by identifying\ndifficult paragraphs using eye-tracking implicitly and generating compre-\nhension questions with a Large Language Model (LLM). Our aim is to\nevaluate the potential of generative AI as a tool to augment human cog-\ngnitive capabilities. If readers answer the questions correctly, this confirms\nthat they understand the text. Conversely, if they answer incorrectly, it\nindicates a lack of understanding, allowing the system to provide explana-\ntions for the misunderstood parts, thereby improving the readers’ under-\nstanding of the text. Our experiment showed that half of the participants\nimproved their comprehension of the text. Further analysis revealed that\nthe motivation to read/use system is one of the most important factor to\nincrease the effectiveness of the proposed system."}},{"key":"Ohyama2025","id":"1979","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"K. Ohyama","en":""},{"ja":"T. Tsukamoto","en":""},{"ja":"T. Yamamoto","en":""},{"ja":"Y. Utsumi","en":""},{"ja":"K. Yoshinaga","en":""},{"ja":"R. Sakata","en":""},{"ja":"S. Tanaka","en":""},{"ja":"T. Isobe","en":""},{"ja":"Y. Miwa","en":""},{"ja":"Y. Suehiro","en":""}],"title":{"ja":"Multipurpose small collaborative robot for smart viticulture: Concept and design"},"journal":{"ja":"Acta Horticulturae"},"volume":"1418","num_pages":"165--172","year":2025,"month":3,"review":"reviewed","fields":["plant-image-processing"],"doi":"10.17660/ActaHortic.2025.1418.21","abstract":{"ja":"in printing"}},{"key":"Islam2025","id":"2036","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Md. Rabiul Islam","en":""},{"ja":"Andrew W. Vargo","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Contrastive Learning for Fine-Grained Reading Detection"},"journal":{"ja":"Information Engineering Express"},"volume":"11","number":"1","year":2025,"month":3,"publisher":"International Institute of Applied Informatics","review":"reviewed","fields":["activity-recognition","deep-learning"],"doi":"10.52731/iee.v11.i1.816","abstract":{"ja":"Reading is a cognitive activity that we perform aiming at various purposes, such as gaining knowledge and entertaining ourselves, with different scripts and layouts. Therefore, automatic reading detection gives useful information about users’ reading activities. Deep learning enables automatic feature extraction and model creation but needs large-sized labeled data. The self-supervised learning devised to overcome this limitation work as noncontrastive self-supervised learning (SSL) and contrastive self-supervised learning (contrastive learning). Although SSL is well explored for reading analysis, contrastive learning is not still well explored. This paper explores contrastive learning that works in several ways. A Simple Framework for Contrastive Learning of Visual Representations (SimCLR) is one way that has attracted much attention in many research domains because of its superior performance. We explore SimCLR for the cognitive activity recognition task of finegrained reading detection employing electrooculography datasets. These datasets describe eye movements that have been recorded for in-the-wild condition. The obtained results are compared against SSL and supervised baselines. The results show that, for an equal number of training samples, the SimCLR method obtains a maximum performance gain of 3.02 and 3.96 percentage points compared to the two baselines, respectively. Besides, SimCLR shows the best performance for large-sized data with a data efficiency of about 80%, whereas SSL shows the best performance for small-sized data. The analysis conducted in this paper shows a direction for researchers and system designers to employ self-supervised learning for automatic reading detection."}},{"key":"Furuya2025","id":"2008","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Mai Furuya","en":""},{"ja":"Tomoya Nakatani","en":""},{"ja":"Yuzuko Utsumi","en":""},{"ja":"Atsushi Matsumura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Morphological trait evaluation of soybean root systems at an early growth stage using image analysis"},"journal":{"ja":"Plant Root"},"volume":"19","pages":"1--12","year":2025,"month":3,"review":"reviewed","fields":["plant-image-processing"],"doi":"10.3117/plantroot.19.1","url":"https://www.jstage.jst.go.jp/article/plantroot/19/0/19_1/_article/-char/en"},{"key":"Selby2025","id":"2015","type":"journal_paper","language":"EN","show":"show_except_datafile","authors":[{"ja":"David Selby","en":""},{"ja":"Yuichiro Iwashita","en":""},{"ja":"Kai Spriestersbach","en":""},{"ja":"Mohammad Saad","en":""},{"ja":"Dennis Bappert","en":""},{"ja":"Archana Warrier","en":""},{"ja":"Sumantrak Mukherjee","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Sebastian Vollmer","en":""}],"title":{"ja":"Had enough of experts? Quantitative knowledge retrieval from large language models"},"journal":{"ja":"Stat"},"volume":"14","number":"2","pages":"e70054","year":2025,"month":3,"publisher":"Wiley","review":"reviewed","fields":["information-retrieval"],"doi":"10.1002/sta4.70054","url":"https://onlinelibrary.wiley.com/doi/10.1002/sta4.70054"},{"key":"Tamura2025","id":"2014","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Yasuto Tamura","en":""},{"ja":"Yuzuko Utsumi","en":""},{"ja":"Yuka Miwa","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Unsupervised monocular depth estimation with omnidirectional camera for 3D reconstruction of grape berries in the wild"},"journal":{"ja":"PLOS ONE"},"volume":"20","number":"2","presen_id":"e0317359","num_pages":"18","year":2025,"month":2,"review":"reviewed","fields":["plant-image-processing","3d-reconstruction","deep-learning"],"doi":"10.1371/journal.pone.0317359","url":"https://doi.org/10.1371/journal.pone.0317359"},{"key":"Ogawa2025","id":"2066","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Tomoya Ogawa","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Soichiro Nakako","en":""},{"ja":"Hiroshi Okamura","en":""},{"ja":"Akinori Nishikawa","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Two-Stage Modeling for Dynamic Survival Prediction from Longitudinal Data"},"journal":{"ja":"NeurIPS 2025 Workshop Learning from Time Series for Health"},"num_pages":"5","year":2025,"month":12,"location":"San Diego, CA, USA","review":"reviewed","presen_type":"poster","fields":[],"url":"https://openreview.net/forum?id=jDPp2lZLvY","abstract":{"ja":"Predicting a patient’s future health risk is essential for medical care, especially when deciding treatments or identifying patients who may need urgent attention. However, many patients’ conditions change over time, meaning that predictions made at one point may become outdated. Dynamic prediction methods address this challenge by updating risk estimates as new laboratory measurements become available. A widely used method called landmarking predicts the risk of an event (such as relapse or death) based on clinical data observed at a specific time point. However, its accuracy often becomes worse as the prediction window becomes longer---that is, the farther into the future we try to predict, the less reliable the prediction becomes. To address this limitation, we propose a simple but effective extension of landmarking called a two-stage modeling approach. Instead of predicting the outcome far into the future all at once, our method first forecasts near-future laboratory measurements and then uses those predicted values to make a shorter-range survival prediction. Even this straightforward naive two-stage method improves accuracy compared with standard landmarking. We further extend this approach by also using summary information about the uncertainty of the laboratory forecast (such as predicted mean and variance). This extended two-stage model achieves better. Across multiple clinical datasets, the proposed two-stage approach consistently improves prediction accuracy over conventional landmarking. These results show that a simple forecasting step can meaningfully improve dynamic risk prediction, offering a practical direction for clinical machine learning models that use longitudinal laboratory values."}},{"key":"Takaishi2025_ASSETSa","id":"2060","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Kosei Takaishi","en":""},{"ja":"Kazunori Minatani","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Disassembling Depth-Estimation-Based 2.5D Photographic Reliefs to Improve Accessibility for People with Visual Impairments"},"journal":{"ja":"Proceedings of the 27th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS 2025)"},"year":2025,"month":10,"location":"Denver, Colorado, USA","review":"reviewed","presen_type":"poster","fields":["visually-impaired-assistance"],"doi":"10.1145/3663547.3759737"},{"key":"Takaishi2025_ASSETSb","id":"2061","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Kosei Takaishi","en":""},{"ja":"Kazunori Minatani","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"A Depth- and Luminance-Based 2.5D Relief to Expand the Expressive Potential of Tactile Photography for People with Visual Impairments"},"journal":{"ja":"Proceedings of the 27th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS 2025)"},"year":2025,"month":10,"location":"Denver, Colorado, USA","review":"reviewed","presen_type":"poster","fields":["visually-impaired-assistance"],"doi":"10.1145/3663547.3759751"},{"key":"Buoy2025_ICDAR","id":"2059","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Rina Buoy","en":""},{"ja":"Sovisal Chenda","en":""},{"ja":"Nguonly Taing","en":""},{"ja":"Marry Kong","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Enhancing Khmer-English Machine Translation via Document Analysis Techniques"},"journal":{"ja":"Proc. the ICDAR 2025 Workshop on Documents Analysis of Low-resource Languages"},"year":2025,"month":9,"location":"Wuhan, Hubei, China","review":"reviewed","fields":["document-image-analysis","deep-learning"]},{"key":"Buoy2025_CITA","id":"2035","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Rina Buoy","en":""},{"ja":"Sovisal Chenda","en":""},{"ja":"Nguonly Taing","en":""},{"ja":"Marry Kong","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Fine-tuning Multilingual Khmer Neural Machine Translation"},"journal":{"ja":"Proc. the 14th Conference on Information Technology and Its Applications (CITA 2025)"},"year":2025,"month":7,"review":"reviewed","fields":["pattern-recognition"]},{"key":"Minatani2025","id":"2048","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Kazunori Minatani","en":""},{"ja":"Tetsuya Watanabe","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Online symposium with tactile 3D model: its general efficacy as a media for the visually impaired"},"journal":{"ja":"Proc. of the 2025 PErvasive Technologies Related to Assistive Environments (PETRA) Conference"},"num_pages":"6","year":2025,"month":6,"review":"reviewed","fields":["visually-impaired-assistance"],"doi":"10.1145/3733155.3733214"},{"key":"Fujii2025","id":"2034","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Haruya Fujii","en":""},{"ja":"Andrew Vargo","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Evaluating In-situ Cognitive Load Using Physiological Data"},"journal":{"ja":"Augmented Humans 2025 (AHs25)"},"year":2025,"month":3,"location":"Abu Dhabi, United Arab Emirates","review":"reviewed","presen_type":"poster","fields":["wellbeing","sensing"],"abstract":{"ja":"In this work, we aim to quantify acute stress induced by cognitive load using real-time physiological data collected by a wearable health-tracking device. In particular, we look at stress induced by increased cognitive load during a series of N-back tasks. Twelve participants completed an N-back task while wearing the Embrace PLUS device, which recorded electrodermal activity (EDA), heart rate variability (HRV), respiratory rate, etc. A linear mixed-effects model revealed significant stress responses, with HR fluctuations as strong predictors. The findings support real-time stress monitoring and adaptive intervention strategies with the use of a werable physiological sensing device."}},{"key":"Hara2025_AHs","id":"2030","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Toma Hara","en":""},{"ja":"Andrew Vargo","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Using wearables to measure the impact of sleep on decision making"},"journal":{"ja":"Augmented Humans 2025 (AHs25)"},"year":2025,"month":3,"location":"Abu Dhabi, United Arab Emirates","review":"reviewed","presen_type":"poster","fields":["wellbeing","sensing"],"abstract":{"ja":"This study examined the link between real-world sleep data, collected via the Oura Ring from 10 students, and their performance on a risk-taking task (BART). The key finding is that average heart rate (HR) and heart rate variability (HRV) during sleep significantly influenced BART outcomes. Specifically, a lower standard deviation of average HR and a higher standard deviation of average HRV over the three days before the task were associated with better risk selection optimization. This suggests that sleep, measured in natural settings, can affect decision-making under uncertainty. However, the study's small sample size limits the generalizability of these findings."}},{"key":"Neigel2025","id":"2046","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Peter Neigel","en":""},{"ja":"David Antony Selby","en":""},{"ja":"Shota Arai","en":""},{"ja":"Benjamin Tag","en":""},{"ja":"Niels van Berkel","en":""},{"ja":"Sebastian Vollmer","en":""},{"ja":"Andrew Vargo","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Exploring the Alignment of Perceived and Measured Sleep Quality with Working Memory Using Consumer Wearables"},"journal":{"ja":"Augmented Humans 2025 (AHs25)"},"year":2025,"month":3,"location":"Abu Dhabi","review":"not_reviewed","presen_type":"oral","fields":["wellbeing","sensing"],"doi":"10.1145/3745900.3746104"},{"key":"Takaishi2025_Symp","id":"2069","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"高石 康世","en":""},{"ja":"南谷 和範","en":""},{"ja":"岩村 雅一","en":""}],"title":{"ja":"視覚障害者による写真鑑賞のための深度推定に基づく2.5Dレリーフの生成と編集"},"journal":{"ja":"感覚代行シンポジウム"},"year":2025,"month":12,"review":"not_reviewed","presen_type":"oral","fields":["visually-impaired-assistance"],"url":"https://www.sensory-substitution.gr.jp/cgi-bin/abst.cgi?vol=051&num=003","awards":{"ja":["2025年感覚代行研究奨励賞受賞"]}},{"key":"Hino2025","id":"2067","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"日野 正大","en":"Shota Hino"},{"ja":"田川 直弥","en":"Tagawa Naoya"},{"ja":"南谷 和範","en":"Kazunori Minatani"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"モジュール化による視覚障害者の3Dプリンタ操作支援システムの汎用化","en":"Generalization of Display Information Recognition Systems for People with Visual Impairments through Modularization"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"125","number":"267","series":"WIT","presen_id":"WIT2025-12","pages":"17-22","year":2025,"month":12,"location":"産総研臨海副都心センター","review":"not_reviewed","presen_type":"oral","fields":["visually-impaired-assistance"],"url":"https://ken.ieice.org/ken/paper/202512036cpY/","awards":{"ja":["WIT学生研究奨励賞受賞"]},"abstract":{"ja":"視覚障害者は，家電などの機器のディスプレイに表示された視覚情報へのアクセスに困難を抱える．そのため，文字情報を視覚障害者に提供するための読み上げシステムが開発されてきた．既存手法は，ディスプレイに表示された文字の読み上げはできるものの，メニューの操作やスクロールする文字列に対応していないため，視覚障害のある利用者にとって，そのような操作は困難であった．\nこの問題に対して田川らは，メニューの操作やスクロールする文字列に対応した読み上げ手法を提案したが，特定の機器のみを対象としており，他の機種への移植性に課題があった．\nそこで本研究では，将来的にさまざまな家電製品に応用可能な読み上げシステムの構築を目指し，その第一歩として複数の3Dプリンタに対応可能なシステム設計に取り組む．従来の読み上げシステムが特定デバイスに依存していた問題を解消するため，認識および読み上げ機能をモジュール化し，異なる機種間で切り替え可能な構成とした．","en":"People with visual impairments face difficulties in accessing visual information displayed on the screens of household appliances and other devices. To address this issue, reading systems have been developed to provide textual information through speech output. However, existing methods can only read aloud the characters displayed on a screen and do not support operations such as navigating menus or handling scrolling text, making such interactions challenging for users with visual impairments.\nTo tackle this problem, Tagawa et al. proposed a reading method capable of handling menu operations and scrolling text. Nevertheless, their approach was limited to a specific device, raising issues of portability to other devices.\nIn this study, we aim to develop a reading system that can be applied to various household appliances in the future. As a first step, we focus on designing a system compatible with multiple 3D printers. To overcome the device-dependency of conventional systems, we modularized the recognition and speech components, enabling flexible switching between different device models."}},{"key":"Takaishi2025_TR","id":"2068","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"高石 康世","en":"Kosei Takaishi"},{"ja":"南谷 和範","en":"Kazunori Minatani"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"深度と輝度に基づく2.5Dレリーフによる視覚障害者向け触覚写真の表現力拡張","en":"Expanding the Expressive Potential of Tactile Photography for People with Visual Impairments through a Depth- and Luminance-Based 2.5D Relief"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"125","number":"267","pages":"5-10","num_pages":"6","year":2025,"month":12,"review":"not_reviewed","presen_type":"oral","fields":["visually-impaired-assistance"],"url":"https://ken.ieice.org/ken/paper/20251203BcPr/","abstract":{"ja":"写真は物語を伝え，記憶を保存し，自己表現するための強力な手段である．しかし，視覚障害者は写真コンテンツへのアクセスに大きな困難を抱えている．最近では Be My AI のような生成AIツールにより画像の言語説明へのアクセスが容易になりつつあるが，言語情報だけでは写真の持つ豊かな情報を十分に伝えることは困難である．\n触覚的な表現がその代替手段として注目されるが，従来の触図（通常は2Dの輪郭に基づくもの）は，表現力に限界がある．そこで本研究は，深度推定と輝度情報の両方を統合した写真からの2.5Dレリーフ生成手法を提案する．この手法により，奥行きを連続的な高さとして豊かに表現しつつ，文字や模様などの輝度値に基づく特徴も凹凸として再現でき，より豊かな触覚的表現を実現する．\n私たちは，視覚障害のある12名の参加者（全盲6名，弱視6名）を対象にユーザスタディを実施した．比較したのは，?2Dの触覚写真，?深度のみの2.5Dレリーフ，?深度と輝度を組み合わせた2.5Dレリーフの3種類の触覚モデルである．また，ユーザの好みを分析するために，5つの評価基準に基づいて階層分析法（AHP: Analytic Hierarchy Process）を適用した．\nその結果，深度情報と輝度情報を組み合わせることの有効性が示され，さらに，写真に触れる際のユーザの動機や経験が触察の仕方にどのように影響をもたらすかについて考察した．","en":"Photographs are powerful tools for storytelling, preserving memories, and self-expression. However, people with visual impairments (PVI) face significant challenges in accessing photographic content.\nRecently, generative AI tools such as Be My AI have made it easier to obtain verbal image descriptions, yet language alone cannot fully convey the rich visual information that photographs contain.\nTactile representation offers an alternative means of access, but traditional tactile graphics?typically based on 2D contours?are limited in expressiveness.\nTo address this limitation, we propose a novel method for generating 2.5D reliefs from photographs by integrating depth estimation and luminance information. This approach allows continuous representation of depth while reproducing luminance-based features such as text and patterns as tactile variations, thereby achieving richer tactile expression.\nWe conducted a user study with 12 participants with visual impairments (six who are blind and six with low vision), comparing three types of tactile models:\n?2D tactile photographs, ?depth-based 2.5D reliefs, and ?depth?luminance-based 2.5D reliefs.\nTo analyze user preferences, we applied the Analytic Hierarchy Process (AHP) based on five evaluation criteria.\nThe results demonstrated the effectiveness of integrating depth and luminance information, and further provided insights into how users' motivation and prior experience influence their tactile exploration and understanding of photographs."}},{"key":"Yodoshi2025","id":"2065","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"吉年 杏将","en":"Kyosuke Yodoshi"},{"ja":"高橋 敬","en":"Kei Takahashi"},{"ja":"南谷 和範","en":"Kazunori Minatani"},{"ja":"渡辺 哲也","en":"Tetsuya Watanabe"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"視覚障害者のための触れる視野ガイドと4台のカメラを用いた3Dプリントモデルの触察支援","en":"Supporting Tactile Exploration of 3D Printed Models Using A Tactile Field-of-View Guide and Four Cameras for People with Visual Impairments"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"125","number":"267","presen_id":"WIT2025-11","pages":"11-16","year":2025,"month":12,"review":"not_reviewed","presen_type":"oral","fields":["visually-impaired-assistance"],"url":"https://ken.ieice.org/ken/paper/20251203vcpX/","abstract":{"ja":"視覚障害者は，写真や図表といった視覚的表現へのアクセスが困難であるため，触覚探索による形状および形態の理解は不可欠である．触覚グラフィックスや3Dプリントモデルは情報へのアクセスにおける代替手段を提供するものの，触覚探索のみでは包括的な理解を得ることが困難な場合が多い．先行研究であるTalkitは，スマートフォンとARマーカーを用いて，ユーザーが3Dモデル上で触れている領域に対する音声説明を提供することを可能にした．しかし，Talkitには2つの重大な制約が存在する．すなわち，3Dモデルがカメラの視野からはずれてしまうこと，および探索中にマーカーが隠れることにより，フィードバックの喪失とユーザーの混乱を招くという問題である．本研究では，折りたたみ式三脚に4台のカメラを設置することで，これらの課題に対処した改良システムを提案する．本システムにより，ユーザーは三脚の脚部に触れることでカメラの視野範囲を認識でき，また複数の視点から3Dモデルおよびユーザーの指先を捉えることで死角を低減できる．本システムは，3Dプリントモデルのアクセシブルな探索を目的としたタッチベース音声注釈システムにおいて，堅牢性および操作性の向上を実現している．","en":"Understanding shapes and forms through tactile exploration is essential for people with visual impairments, as visual representations such as photographs and diagrams are often inaccessible. While tactile graphics and 3D-printed models offer alternative means to access such information, tactile exploration alone is often insufficient for comprehensive understanding. Prior work, such as Talkit, enables users to receive verbal descriptions of the regions they are touching on 3D models using a smartphone and AR markers. However, Talkit suffers from two major limitations: the 3D model may move out of the camera's view, and the markers may be occluded during exploration, leading to loss of feedback and user confusion. In this research, we present an improved system that addresses these issues by mounting four cameras on a collapsible tripod. This configuration allows users to remain aware of the camera's field of view by touching the tripod legs and reduces blind spots by capturing the model and the user's fingers from multiple angles. The system improves the robustness and usability of touch-based verbal annotation systems for accessible exploration of 3D-printed objects."}},{"key":"Yamazaki2025_Proc","id":"2081","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"山崎 匠真","en":""},{"ja":"梁 誠","en":""},{"ja":"田所 晃汰","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"三輪 由佳","en":""},{"ja":"岩村 雅一","en":""}],"title":{"ja":"3次元ブドウ房再構成に基づく自動摘粒選択"},"journal":{"ja":"映像情報メディア学会 関西支部 若手研究発表会 予稿集"},"num_pages":"2","year":2025,"month":11,"review":"not_reviewed","presen_type":"oral","fields":["3d-reconstruction","plant-image-processing"],"awards":{"ja":["優秀論文発表賞受賞"]}},{"key":"Iwashita2025_Conf","id":"2118","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩下 雄一郎","en":""},{"ja":"筒崎 乙華","en":""},{"ja":"落合 梓香","en":""},{"ja":"井上 達意","en":""},{"ja":"野々宮 悠太","en":""},{"ja":"光山 容仁","en":""},{"ja":"仲子 聡一郎","en":""},{"ja":"岩村 雅一","en":""},{"ja":"Muhammad Nabeel Asim","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"マルチモーダルEHRを用いた動的予測ベンチマークの構築"},"journal":{"ja":"第45回医療情報学連合大会（第26回日本医療情報学会学術大会）"},"presen_id":"4-E-1-01","year":2025,"month":11,"location":"兵庫県姫路市","review":"reviewed","presen_type":"oral","fields":["medical-ai"],"url":"https://pub.confit.atlas.jp/ja/event/jcmi2025/presentation/4-E-1-01"},{"key":"Ogawa2025_Conf","id":"2117","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"小川 智也","en":"Tomoya Ogawa"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"仲子 聡一郎","en":"Soichiro Nakako"},{"ja":"岡村 浩史","en":"Hiroshi Okamura"},{"ja":"西川 彰則","en":"Akinori Nishikawa"},{"ja":"森口 慎","en":"Makoto Moriguchi"},{"ja":"中舎 洋輔","en":"Yosuke Nakaya"},{"ja":"堀内 美令","en":"Mirei Horiuchi"},{"ja":"長? 譲慈","en":"Joji Nagasaki"},{"ja":"井戸 健太郎","en":"Kentaro Ido"},{"ja":"久野 雅智","en":"Masatomo Kuno"},{"ja":"西本 光孝","en":"Mitsutaka Nishimoto"},{"ja":"中嶋 康博","en":"Yasuhiro Nakashima"},{"ja":"中前 美佳","en":"Mika Nakamae"},{"ja":"中前 博久","en":"Hirohisa Nakamae"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"2段階の回帰を用いた同種造血幹細胞移植後合併症の動的予測モデル","en":"A Dynamic Prediction Model for Complications Following Allogeneic Hematopoietic Stem Cell Transplantation Using Two-Stage Regression"},"journal":{"ja":"第45回医療情報学連合大会（第26回日本医療情報学会学術大会）"},"presen_id":"4-E-1-03","year":2025,"month":11,"review":"reviewed","presen_type":"oral","fields":["medical-ai"],"url":"https://pub.confit.atlas.jp/ja/event/jcmi2025/presentation/4-E-1-03"},{"key":"Ryuu2025","id":"2098","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"劉 少楠","en":""},{"ja":"岩下 雄一郎","en":""},{"ja":"仲子 聡一郎","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"院内死亡予測のためのTransformerベースの拡散モデルによる合成時系列EHRの生成"},"journal":{"ja":"第45回医療情報学連合大会（第26回日本医療情報学会学術大会）"},"year":2025,"month":11,"review":"reviewed","presen_type":"oral","fields":["medical-ai"]},{"key":"Tsutsusaki2025","id":"2099","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"筒崎 乙華","en":""},{"ja":"岩下 雄一郎","en":""},{"ja":"劉 少楠","en":""},{"ja":"井上 達意","en":""},{"ja":"野々宮 悠太","en":""},{"ja":"仲子 聡一郎","en":""},{"ja":"岡村 浩史","en":""},{"ja":"岩村 雅一","en":""},{"ja":"石丸 翔也","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"マルチモーダルな医療データを用いた ICU 患者の動的予後予測モデルの構築"},"journal":{"ja":"第45回医療情報学連合大会（第26回日本医療情報学会学術大会）"},"year":2025,"month":11,"review":"reviewed","presen_type":"oral","fields":["medical-ai"]},{"key":"Yamazaki2025_MIRU","id":"2082","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"山崎匠真","en":""},{"ja":"内海ゆづ子","en":""},{"ja":"三輪由佳","en":""},{"ja":"岩村雅一","en":""}],"title":{"ja":"ブドウの3 次元構造把握のための球体検出を用いた果粒検出"},"journal":{"ja":"画像の認識・理解シンポジウム（MIRU）予稿集"},"presen_id":"IS3-004","pages":"pp.1-5","num_pages":"5","year":2025,"month":8,"review":"not_reviewed","presen_type":"poster","fields":["3d-reconstruction","plant-image-processing"]},{"key":"Sudou2025_MIRU","id":"2096","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"須藤友貴","en":""},{"ja":"森山一斗","en":""},{"ja":"内海ゆづ子","en":""},{"ja":"松村篤","en":""},{"ja":"岩村雅一","en":""}],"title":{"ja":"遺伝要因解明のための楕円フーリエ解析を用いたダイズ根の形状解析"},"journal":{"ja":"第28回 画像の認識・理解シンポジウム (MIRU2025)予稿集"},"presen_id":"IS1-172","num_pages":"5","year":2025,"month":8,"review":"not_reviewed","presen_type":"poster","fields":["pattern-recognition","plant-image-processing"]},{"key":"Wasaka2025_MIRU","id":"2088","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"和坂優佑","en":""},{"ja":"内海ゆづ子","en":""},{"ja":"三輪由佳","en":""},{"ja":"岩村雅一","en":""}],"title":{"ja":"全方位カメラ画像を用いたブドウの房計数手法におけるドメイン一般化の有効性の検証"},"journal":{"ja":"第28回画像の認識・理解シンポジウム（MIRU2025）予稿集"},"presen_id":"IS1-173","num_pages":"5","year":2025,"month":8,"review":"not_reviewed","presen_type":"poster","fields":["plant-image-processing","pattern-recognition"]},{"key":"Kawagoe2025_CVIM","id":"2056","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"川越壮","en":""},{"ja":"平野甫","en":""},{"ja":"岩村雅一","en":""}],"title":{"ja":"LLM説明文とQ-Formerの活用によるCLIP Few-shot Adaptationの精度向上"},"journal":{"ja":"情報処理学会 研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2025-CVIM-242","presen_id":"50","pages":"1-8","num_pages":"8","year":2025,"month":5,"publisher":"情報処理学会","location":"奈良女子大学","review":"not_reviewed","presen_type":"poster","fields":["deep-learning","object-recognition"],"url":"https://ipsj.ixsq.nii.ac.jp/records/2001881"},{"key":"Shinoda2025_CVIM","id":"2057","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"信田 浩希","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"藤本 仰一","en":""},{"ja":"岩村 雅一","en":""}],"title":{"ja":"花弁配置推定システムのための合成画像を用いた分類手法の評価"},"journal":{"ja":"情報処理学会 研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2025-CVIM-242","presen_id":"32","pages":"1-8","num_pages":"8","year":2025,"month":5,"publisher":"情報処理学会","location":"奈良女子大学","review":"not_reviewed","presen_type":"poster","fields":["plant-image-processing","object-recognition"],"url":"https://ipsj.ixsq.nii.ac.jp/records/2001863","abstract":{"ja":"花弁配置は, 花弁の枚数とその位置, それらの前後関係を表すものであり, 種を超えた原理に基づくと考えられている. 数理モデルにより, 花弁の発生過程のシミュレーションが行われており, 数理モデルの考案や検証のために, 画像から花弁配置を自動推定するシステムが求められている. 従来手法では花弁の重なり位置の検出と前後関係の推定を別々に行い, 推定精度が0.275と低かった. 本研究では, Conditional DETRを用いて両者を統合的に処理し, 精度向上を試みた. 合成画像を用いた実験の結果, 推定精度は0.343となった."}},{"key":"Sudou2025_CVIM","id":"2051","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"須藤 友貴","en":""},{"ja":"森山 一斗","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"松村 篤","en":""},{"ja":"岩村 雅一","en":""}],"title":{"ja":"遺伝要因解明のための楕円フーリエ解析を用いた圃場のダイズ根の形状解析"},"journal":{"ja":"情報処理学会 研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2025-CVIM-242","presen_id":"14","pages":"1-8","num_pages":"8","year":2025,"month":5,"publisher":"情報処理学会","location":"奈良女子大学","review":"not_reviewed","presen_type":"poster","fields":["plant-image-processing","object-recognition"],"url":"https://ipsj.ixsq.nii.ac.jp/records/2001845","abstract":{"ja":"植物の根の形状は植物の栄養獲得戦略に大きく影響を与えるため，生育環境に合う形状の根をもつ植 物の育種を実現することが望まれている．効率的な育種は遺伝子操作であるが，根形状の関連遺伝子の特定 が必要である．近年，ゲノムワイド関連解析(Genome-Wide Association Studies：GWAS) で遺伝子の特 定が容易になったが，GWAS の適用には根の形状を定量化する必要がある．根形状を定量化した既存の研 究は屋内栽培のダイズを用いており，通常，ダイズが生育される圃場との環境差により根形状が異なるた め，解析手法や結果は実用的でない．そこで，本研究では圃場で栽培されたダイズ根画像から根の概形を 抽出し，楕円フーリエ解析を用いて形状の定量化を行う．本研究で得たダイズ根の形状の特徴量をゲノム ワイド関連解析で遺伝子との関連を調べた結果，形状の特徴量と関連する遺伝子の変異が一つ検出された．"}},{"key":"Wasaka2025_CVIM","id":"2055","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"和坂 優佑","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"三輪 由佳","en":""},{"ja":"岩村 雅一","en":""}],"title":{"ja":"異なる撮影環境に対応したブドウの房計数手法の検証"},"journal":{"ja":"情報処理学会 研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2025-CVIM-242","number":"18","pages":"1-8","num_pages":"8","year":2025,"month":5,"location":"奈良女子大学","review":"not_reviewed","presen_type":"poster","fields":["plant-image-processing","object-recognition"],"url":"https://ipsj.ixsq.nii.ac.jp/records/2001849","abstract":{"ja":"ブドウ栽培における摘房は，単位面積あたりのブドウの房の数が一定の数になるように房を間引く作業である．現状の摘房作業では，ブドウの房の数は正確には数えておらず，作業者の経験や感覚で摘房している．そこで本研究では正確な房数の把握を目的とし，全方位カメラで撮影した画像を用いた房の計数システムを構築する．既存研究では，撮影環境が変化した場合の計数モデルの精度低下が課題であった．そのため本研究ではドメイン一般化手法を適用し，撮影環境に依存しない計数の実現を目指す．撮影した年が異なる2つのデータセットを用いて実験した結果，一部の結果では異なるデータセットに対しても精度を維持できたが，全体的に実用化できるほどの精度に達していないことが分かった．また撮影時の全方位カメラとブドウ棚の距離が計数精度に影響を与えていることが示唆された．"}},{"key":"Arai2025","id":"2044","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"新井将太","en":""},{"ja":"Andrew Vargo","en":""},{"ja":"Peter Neigel","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"一週間の睡眠傾向と作業記憶の関係性の調査"},"journal":{"ja":"インタラクション2025論文集"},"volume":"3P-76","pages":"1342-1347","num_pages":"5","year":2025,"month":2,"publisher":"IPSJ","location":"Tokyo","review":"not_reviewed","fields":["wellbeing"],"url":"https://www.interaction-ipsj.org/proceedings/2025/data/bib/3P-76.html"},{"key":"Nolasco2025_Sleep","id":"2045","type":"domestic_conf","language":"EN","show":"show","authors":[{"ja":"Hannah R. Nolasco","en":""},{"ja":"Andrew Vargo","en":""},{"ja":"Ko Watanabe","en":""},{"ja":"Koichi Kise","en":""}],"title":{"en":"Sleep Tracking: The False Dichotomy between ManualMethods and Automation with Wearables"},"journal":{"ja":"インタラクション2025論文集"},"volume":"3P-78","pages":"1354-1358","year":2025,"month":2,"review":"not_reviewed","fields":[],"url":"https://www.interaction-ipsj.org/proceedings/2025/data/pdf/3P-78.pdf"},{"key":"Otsuka2025","id":"2039","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"大塚 遙","en":""},{"ja":"岩田 基","en":""},{"ja":"岩下 雄一郎","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"生成AIで画像を作りながら学べる英単語学習システム"},"journal":{"ja":"インタラクション2025論文集"},"presen_id":"1B-28","pages":"348-353","num_pages":"6","year":2025,"month":2,"publisher":"情報処理学会","location":"学術総合センター内 一橋記念講堂","review":"not_reviewed","presen_type":"poster","fields":["learning-analytics"]},{"key":"Kakei2025","id":"2041","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"筧 万里","en":""},{"ja":"岩村 雅一","en":""},{"ja":"南谷 和範","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"視覚障害者のための棒グラフのグループ化による要約"},"journal":{"ja":"情報アクセシビリティをめぐる諸問題に関する研究集会"},"year":2025,"month":2,"review":"not_reviewed","presen_type":"oral","fields":["visually-impaired-assistance"]},{"key":"Tagawa2025","id":"2042","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"田川 直弥","en":""},{"ja":"岩村 雅一","en":""},{"ja":"南谷 和範","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"メニューとスクロールする文字列の読み上げによる視覚障害者の3Dプリンタのアクセシビリティ向上"},"journal":{"ja":"情報アクセシビリティをめぐる諸問題に関する研究集会"},"year":2025,"month":2,"review":"not_reviewed","presen_type":"oral","fields":["visually-impaired-assistance"]},{"key":"Seki2025","id":"2022","type":"bachelors_thesis","language":"EN","show":"show","authors":[{"ja":"関勇人","en":"Hayato Seki"}],"title":{"ja":"日本人とフランス人における英語漫画音読時の未知単語推定の性能比較","en":"Comparison of unknown word estimation performance between Japanese and French people when reading English comics aloud"},"num_pages":"25","year":2025,"month":2,"review":"not_reviewed","fields":["sensing","activity-recognition"]},{"key":"Nolasco2025_Solutionism","id":"2114","type":"other","language":"EN","show":"show","authors":[{"ja":"","en":"Hannah R. Nolasco"},{"ja":"","en":"Andrew Vargo"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"AI Solutionism and Digital Self-Tracking with Wearables"},"year":2025,"month":5,"review":"not_reviewed","fields":["wellbeing"],"url":"https://doi.org/10.48550/arXiv.2505.15162"},{"key":"Chowdhury2025","id":"2115","type":"other","language":"EN","show":"show","authors":[{"ja":"","en":"Tasnim Irtifa Chowdhury"},{"ja":"","en":"Andrew Vargo"},{"ja":"","en":"Chris Blakely"},{"ja":"","en":"Benjamin Tag"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"A Review of Cognitive Readiness, Wearable Devices, and Prospects"},"journal":{"en":"arXiv"},"year":2025,"month":1,"review":"not_reviewed","fields":["wellbeing"],"url":"https://doi.org/10.48550/arXiv.2501.03537"},{"key":"Maruyama2024","id":"2007","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Mizuki Maruyama","en":""},{"ja":"Shrey Singh","en":""},{"ja":"Katsufumi Inoue","en":""},{"ja":"Partha Pratim Roy","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Michifumi Yoshioka","en":""}],"title":{"ja":"Word-level Sign Language Recognition with Multi-stream Neural Networks Focusing on Local Regions and Skeletal Information"},"journal":{"ja":"IEEE ACCESS"},"volume":"12","pages":"167333-167346","year":2024,"month":11,"publisher":"IEEE","review":"reviewed","fields":["sign-language-recognition"],"doi":"10.1109/ACCESS.2024.3494878"},{"key":"Higashimura2024_IEEE","id":"2002","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Riku Higashimura","en":""},{"ja":"Ko Watanabe","en":""},{"ja":"Andrew Vargo","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Estimating Unknown English Words From User Smartphone Reading Behaviors"},"journal":{"ja":"IEEE Access"},"num_pages":"14","year":2024,"month":9,"publisher":"IEEE","review":"reviewed","fields":["learning-analytics","sensing"],"doi":"10.1109/ACCESS.2024.3457510"},{"key":"Matsumoto2024","id":"1990","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Soushi Matsumoto","en":""},{"ja":"Yuzuko Utsumi","en":""},{"ja":"Toshiaki Kozuka","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Tomonori Nakai","en":""},{"ja":"Daisuke Yamauchi","en":""},{"ja":"Ichirou Karahara","en":""},{"ja":"Yoshinobu Mineyuki","en":""},{"ja":"Masato Hoshino","en":""},{"ja":"Kentaro Uesugi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"CT Image-based 3D Inflorescence Estimation of Chrysanthemum seticuspe"},"journal":{"ja":"Frontiers in Plant Science"},"volume":"15","number":"1374937","year":2024,"month":7,"review":"reviewed","fields":["plant-image-processing"],"doi":"10.3389/fpls.2024.1374937"},{"key":"Naka2024","id":"1989","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Yuki Naka","en":""},{"ja":"Yuzuko Utsumi","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Hirokazu Tsukaya","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Petal Segmentation in CT Images Based on Divide-and-Conquer Strategy"},"journal":{"ja":"Frontiers in Plant Science"},"volume":"15","number":"1389902","year":2024,"month":7,"review":"reviewed","fields":["plant-image-processing"],"doi":"10.3389/fpls.2024.1389902"},{"key":"Buoy2024_IJDAR_Parstr","id":"1976","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"","en":"Rina Buoy"},{"ja":"","en":"Masakazu Iwamura"},{"ja":"","en":"Sovila Srun"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"PARSTR: Partially Autoregressive Scene Text Recognition"},"journal":{"en":"International Journal on Document Analysis and Recognition (IJDAR)"},"volume":"27","pages":"303-316","num_pages":"14","year":2024,"month":5,"publisher":"Springer","address":"Berlin, Germany","review":"reviewed","fields":["document-image-analysis","pattern-recognition","deep-learning"],"doi":"10.1007/s10032-024-00470-1","abstract":{"en":"An autoregressive (AR) decoder for scene text recognition (STR) requires numerous generation steps to decode a text image character by character but can yield high recognition accuracy. On the other hand, a non-autoregressive (NAR) decoder generates all characters in a single generation but suffers from a loss of recognition accuracy. This is because, unlike the former, the latter assumes that the predicted characters are conditionally independent. This paper presents a Partially Autoregressive Scene Text Recognition (PARSTR) method that unifies both AR and NAR decoding within the same model. To reduce decoding steps while maintaining recognition accuracy, we devise two decoding strategies: b-first and b-ahead, reducing the decoding steps to approximately b and by a factor of b, respectively. The experimental results demonstrate that our PARSTR models using the devised decoding strategies present a balanced compromise between efficiency and recognition accuracy compared to the fully AR and NAR decoding approaches. Specifically, the experimental results on public benchmark STR datasets demonstrate the potential to reduce decoding steps down to at most five steps and by a factor of five under the b-first and b-ahead decoding schemes, respectively, while having a slight reduction of total word recognition accuracy of less than or equal to 0.5%."}},{"key":"Buoy2024_IJDAR_Reduced","id":"1977","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"","en":"Rina Buoy"},{"ja":"","en":"Masakazu Iwamura"},{"ja":"","en":"Sovila Srun"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Towards Reduced-Complexity Scene Text Recognition (RCSTR) Through A Novel Salient Feature Selection"},"journal":{"en":"International Journal on Document Analysis and Recognition (IJDAR)"},"volume":"27","pages":"289-302","num_pages":"14","year":2024,"month":5,"publisher":"Springer","address":"Berlin, Germany","review":"reviewed","fields":["document-image-analysis","pattern-recognition","deep-learning"],"doi":"10.1007/s10032-024-00474-x","abstract":{"en":"The integration of an attention mechanism has played a crucial role in many recent scene text recognition (STR) methods. It enables the capture of spatial feature dependencies (known as self-attention) and the identification of relevant features while predicting a character (known as cross-attention). However, computations and memory requirements in the self-attention and cross-attention layers increase quadratically and linearly with the feature map size, respectively, leading to a computational bottleneck in low-resource environments. But, is it necessary to attend to the entire feature maps? On the other hand, text in a natural scene is continuous and oriented in a specific direction, and it does not occupy the entire image. Therefore, utilizing only a small salient subset of features in text regions is sufficient for accurately predicting characters. Based on this salient feature selection, we propose a reduced-complexity scene text recognition framework that significantly reduces model complexities and memory requirements in the self-attention and cross-attention layers. We validate the proposed framework by employing a convolutional STR architecture with both connectionist temporal classification and transformer decoders. Through the model complexity and performance analyses on public benchmark datasets, we demonstrate that the proposed method can substantially reduce model complexities while still maintaining reasonably robust recognition accuracy."}},{"key":"Nakatani2024","id":"1971","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Tomoya Nakatani","en":""},{"ja":"Yuzuko Utsumi","en":""},{"ja":"Koichi Fujimoto","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Image recognition-based petal arrangement estimation"},"journal":{"ja":"Frontiers in Plant Science"},"volume":"15","number":"1334362","num_pages":"14","year":2024,"month":4,"review":"reviewed","fields":["plant-image-processing"],"doi":"10.3389/fpls.2024.1334362"},{"key":"Selby2024","id":"2006","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"David Selby","en":""},{"ja":"Kai Spriestersbach","en":""},{"ja":"Yuichiro Iwashita","en":""},{"ja":"Dennis Bappert","en":""},{"ja":"Archana Warrier","en":""},{"ja":"Sumantrak Mukherjee","en":""},{"ja":"Sergey Redyuk","en":""},{"ja":"Muhammad Nabeel Asim","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Sebastian Vollmer","en":""}],"title":{"ja":"Had Enough of Experts? Elicitation and Evaluation of Bayesian Priors from Large Language Models"},"journal":{"ja":"NeurIPS 2024 Workshop on Bayesian Decision-making and Uncertainty"},"year":2024,"month":12,"location":"Vancouver, Canada","review":"reviewed","presen_type":"poster","fields":["information-retrieval"]},{"key":"Pandey2024","id":"1992","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Vishal Pandey","en":""},{"ja":"Nikhil Panwar","en":""},{"ja":"Atharva Kumbhar","en":""},{"ja":"Partha Pratim Roy","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Enhanced Cross-Task EEG Classification: Domain Adaptation with EEGNet"},"journal":{"ja":"Proc. 27th International Conference on Pattern Recognition (ICPR)"},"pages":"354-369","year":2024,"month":12,"publisher":"Springer","location":"Kolkata, India","review":"reviewed","presen_type":"poster","fields":["sensing","deep-learning"],"doi":"10.1007/978-3-031-78195-7_24"},{"key":"Siddhad2024","id":"1988","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Gourav Siddhad","en":""},{"ja":"Sayantan Dey","en":""},{"ja":"Partha Pratim Roy","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Awake at the Wheel: Enhancing Automotive Safety through EEG-Based Fatigue Detection"},"journal":{"ja":"Proc. 27th International Conference on Pattern Recognition (ICPR)"},"pages":"340-353","year":2024,"month":12,"publisher":"Springer","location":"Kolkata, India","review":"reviewed","presen_type":"oral","fields":["sensing","deep-learning"],"doi":"10.1007/978-3-031-78195-7_23"},{"key":"Chowdhury2024","id":"1997","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Tasnim Irtifa Chowdhury","en":""},{"ja":"Andrew Vargo","en":""},{"ja":"Benjamin Tag","en":""},{"ja":"Ryunosuke Nishitomi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Exploring the Feasibility of a Repeated Mobile One-Minute PVT In-the-Wild"},"journal":{"ja":"MobileHCI '24 Adjunct: Adjunct Proceedings of the 26th International Conference on Mobile Human-Computer Interaction"},"pages":"1-7","num_pages":"7","year":2024,"month":10,"location":"Melbourne, Australia","review":"reviewed","fields":["sensing","wellbeing"],"doi":"10.1145/3640471.3680235","abstract":{"ja":"Researchers have established that quality sleep is important for cognitive functions. Now that consumer-grade wearables can provide accurate measurements of sleep quality, we should be able to inform individuals about how their sleep impacts cognitive functions. However, as our internal clock follows cycles of alertness and sleepiness, it is still necessary to identify at what time of day the sleep quality impacts cognitive performance. In this study, we investigated the relationship between sleep stages and psychomotor vigilance using a 1-minute Psychomotor Vigilance Task (PVT) test. Participants wore a sleep-tracking device and performed the PVT test six times daily for three weeks. The results suggest that increased Rapid Eye Movement (REM) sleep duration led to better performance after lunch. Additionally, we developed a model to predict average response time based on the sleep data. The results show a promising step towards giving tangible meaning to sleep data."}},{"key":"Singh2024","id":"1987","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Shrey Singh","en":""},{"ja":"Prateek Keserwani","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Partha Pratim Roy","en":""}],"title":{"ja":"DCDM: Diffusion-Conditioned-Diffusion Model for Scene Text Image Super-Resolution"},"journal":{"ja":"Proc. 18th European Conference on Computer Vision (ECCV)"},"year":2024,"month":10,"location":"Milan, Italy","review":"reviewed","presen_type":"poster","fields":["document-image-analysis","deep-learning"],"doi":"10.1007/978-3-031-72633-0_17","url":"https://www.ecva.net/papers/eccv_2024/papers_ECCV/html/2357_ECCV_2024_paper.php","note":{"ja":"27.9% acceptance rate at a top conference of computer vision"}},{"key":"Kakehi2024","id":"1994","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Banri Kakehi","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Kazunori Minatani","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Grouping Effect for Bar Graph Summarization for People with Visual Impairments"},"journal":{"ja":"Proceedings of the 26th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS 2024)"},"year":2024,"month":10,"location":"St. John's, NL, Canada","review":"reviewed","presen_type":"poster","fields":["visually-impaired-assistance"],"doi":"10.1145/3663548.3688534","abstract":{"ja":"When communicating numerical data to people with visual impairments (PVI), summaries provided by current data visualization solutions tend to lose important information during summarization. To address this issue, our work focuses on summarization through bar grouping in bar graphs. Neither the effect of grouping nor the appropriate granularity of grouping has been discussed so far. Therefore, we investigate the cognitive effects of grouping and its relationship to the number of groups. A user study involving nine PVI (five blind and four with low vision) revealed that summarization through bar grouping conveys information significantly more accurately compared to simply reading individual data points, despite the inherent error produced by grouping. Additionally, we propose a cognitive error model to explain the characteristics of the observed errors."}},{"key":"Tagawa2024","id":"1993","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Naoya Tagawa","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Kazunori Minatani","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Making 3D Printer Accessible for People with Visual Impairments by Reading Scrolling Text and Menus"},"journal":{"ja":"Proceedings of the 26th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS 2024)"},"num_pages":"4","year":2024,"month":10,"location":"St. John's, NL, Canada","review":"reviewed","presen_type":"poster","fields":["visually-impaired-assistance"],"doi":"10.1145/3663548.3688517","abstract":{"ja":"3D printing has immense potential to enhance the lives of people with visual impairments (PVI) by enabling them to understand shapes and other details through touch that words alone cannot convey. Several initiatives have made 3D tactile models accessible to PVI, yet these models are typically created by sighted individuals. Our goal is to empower PVI to create 3D tactile models independently, making 3D printers accessible to them. The biggest bottleneck for PVI in using 3D printers is the inability to read text on their display. Our work specifically focuses on making scrolling text and menus readable. Through a user study with 13 PVI (five blind and eight with low vision), we confirmed the effectiveness of the implemented functions over conventional smartphone apps and wearable devices."}},{"key":"Arai2024_ISWC","id":"2001","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Shota Arai","en":""},{"ja":"Andrew Vargo","en":""},{"ja":"Benjamin Tag","en":""},{"ja":"Peter Neigel","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"In-the-Wild Exploration of the Impact of Sleep on Working Memory with a Wearable"},"journal":{"ja":"UbiComp/ISWC '24 Adjunct: Adjunct Proceedings of the 2024 ACM International Joint Conference on Pervasive and Ubiquitous Computing & the 2024 ACM International Symposium on Wearable Computing"},"pages":"81-85","num_pages":"5","year":2024,"month":10,"publisher":"ACM","review":"reviewed","fields":["wellbeing","sensing"],"doi":"10.1145/3675094.3677589"},{"key":"Iwashita2024_ISWC","id":"2000","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Yuichiro Iwashita","en":""},{"ja":"Ryuki Kurashige","en":""},{"ja":"Andrew Vargo","en":""},{"ja":"Benjamin Tag","en":""},{"ja":"Tilman Dingler","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Peer-Awareness to Support Learning: An In-the-wild Study on Notification Timing"},"journal":{"ja":"UbiComp/ISWC '24 Adjunct: Adjunct Proceedings of the 2024 ACM International Joint Conference on Pervasive and Ubiquitous Computing & the 2024 ACM International Symposium on Wearable Computing"},"num_pages":"5","year":2024,"month":10,"review":"reviewed","fields":["learning-analytics","sensing"],"doi":"10.1145/3675094.3677576"},{"key":"Watanabe2024","id":"1999","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Ko Watanabe","en":""},{"ja":"Seiya Tanaka","en":""},{"ja":"Andrew Vargo","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Andreas Dengel","en":""}],"title":{"ja":"Comparing Web Browsing Behaviors with High and Low Information Literacy: A Case Study for Fact Check Against GPT Generated Fake News"},"journal":{"ja":"UbiComp/ISWC '24 Adjunct: Adjunct Proceedings of the 2024 ACM International Joint Conference on Pervasive and Ubiquitous Computing & the 2024 ACM International Symposium on Wearable Computing"},"pages":"9-13","year":2024,"month":10,"publisher":"ACM","review":"reviewed","fields":["learning-analytics","sensing"],"url":"https://doi.org/10.1145/3675094.3677575"},{"key":"Iwata2024","id":"2003","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"","en":"Motoi Iwata"},{"ja":"","en":"Keito Okamoto"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Quantitative evaluation based on CLIP for methods inhibiting imitation of painting styles"},"journal":{"en":"LNCS 14935 Document Analysis and Recognition - ICDAR 2024 Workshops"},"pages":"216-230","num_pages":"15","year":2024,"month":8,"publisher":"Springer","location":"Athens, Greece","review":"reviewed","presen_type":"oral","fields":["comic-computing","document-image-analysis","pattern-recognition"],"doi":"10.1007/978-3-031-70645-5_14","url":"https://link.springer.com/chapter/10.1007/978-3-031-70645-5_14"},{"key":"Neigel2024_INTERACT","id":"1986","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Peter Neigel","en":""},{"ja":"Andrew Vargo","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"A Summary of the Workshop on Intelligence Augmentation: Future Directions and Ethical Implications in HCI"},"journal":{"ja":"Design for Equality and Justice; INTERACT 2023 IFIP TC 13 Workshops"},"year":2024,"month":7,"publisher":"Springer","location":"York, UK","review":"reviewed","fields":["sensing"],"doi":"10.1007/978-3-031-61698-3","url":"https://link.springer.com/book/10.1007/978-3-031-61698-3"},{"key":"Buoy2024","id":"1978","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Rina Buoy","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Sovila Srun","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Language-Aware Non-Autoregressive Khmer Textline Recognition Using Khmer Subword Model"},"journal":{"ja":"Proc. International Conference on Pattern Recognition and Artificial Intelligence"},"num_pages":"16","year":2024,"month":7,"location":"Jeju, Korea","review":"reviewed","presen_type":"oral","fields":["document-image-analysis","deep-learning","pattern-recognition"],"abstract":{"ja":"Unlike the Latin script, Khmer does not use spaces between words, leading to text recognition typically being done at the textline level. This can involve a vast number of characters and results in high latency for a language-aware autoregressive (AR) decoder that generates one character at a time. On the other hand, a non-autoregressive (NAR) decoder generates all characters in parallel, but it is not language-aware. In this paper, we introduce an efficient Khmer textline recognition method based on a NAR decoder, ensuring low decoding latency while maintaining linguistic awareness. This is achieved by utilizing a Khmer-specific subword modeling called Khmer character clusters (KCC) that capture the syntactic, morphological, and orthographic aspects of the Khmer script. Therefore, instead of conventional character-level recognition, the proposed method recognizes all character clusters or subwords in parallel. The experimental results demonstrate that the proposed method outperforms the character-level baseline NAR model in terms of recognition accuracy while maintaining the same low latency. When compared with the character-level baseline AR model, the proposed method achieves comparable or improved recognition accuracy while also achieving significantly lower latency. When compared with the recent state-of-the-art (SOTA) NAR and AR Khmer text recognition methods, our proposed method achieves superior recognition performance."}},{"key":"Arai2024","id":"1970","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Shota Arai","en":""},{"ja":"Andrew Vargo","en":""},{"ja":"Benjamin Tag","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"In-the-Wild Exploration of the Impact of the Lunar Cycle on Sleep in a University Cohort with Oura Rings"},"year":2024,"month":4,"publisher":"Augmented Humans 2024 (AHs24)","location":"Melbourne, Australia","review":"reviewed","presen_type":"poster","fields":[],"doi":"10.1145/3652920.3653044","abstract":{"ja":"Adequate sleep is an essential part of human life. The cumulative effects of sleep deprivation and sleep disorders are associated with a wide range of negative health effects and can cause issues with cognitive readiness. Using wearable physiological-sensing devices to monitor sleep and sleep patterns could lead to interventions where individuals and groups can improve their lives. In this study, we analyzed in-the-wild sleep data obtained from Oura Ring, a ring-shaped wearable device specializing in sleep tracking, for 74 university students over a minimum period of 5 months. We investigated the influence of the lunar cycle on sleep and found that the periodicity of the moon affects sleep. Since the lunar cycle has shown to to affect sleep duration and quality, it provides an interesting test-case for in-the-wild data collection form a large cohort. The results of this paper show the potential and limitations of physiological tracking for cohorts in an in-the-wild context."}},{"key":"Iwashita2024","id":"1969","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Yuichiro Iwashita","en":""},{"ja":"Andrew Vargo","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Koichi Kise","en":""}],"title":{"en":"Personalizing Augmented Flashcards Towards Long-Term Vocabulary Learning"},"book_title":"Proceedings of the Augmented Humans International Conference 2024","series":"AHs '24","pages":"263-266","num_pages":"4","year":2024,"month":4,"publisher":"Association for Computing Machinery","address":"New York, NY, USA","location":"Melbourne, VIC, Australia","review":"reviewed","presen_type":"poster","fields":["learning-analytics","sensing","pattern-recognition"],"doi":"10.1145/3652920.3653045","url":"https://doi.org/10.1145/3652920.3653045","abstract":{"ja":"Flashcards are one of the most popular tools for learning vocabulary for second-language learners. While flashcard mediated learning is efficient, it may not induce motivation for continued use. Some researchers have proposed augmented flashcards that provide multimedia contexts to motivate people to study. However, the augmented flashcards also have a problem that it takes time to learn each target word. Understanding this tradeoff, we introduce a system that users can learn vocabulary with both standard and augmented flashcards. In addition, our system recommends the best learning strategy to users adaptively, and realizes the long-term vocabulary learning. In this paper, we describe the system and present the results of the preliminary data analysis towards the long-term vocabulary learning."}},{"key":"Nolasco2024","id":"1967","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Hannah Nolasco","en":""},{"ja":"Andrew Vargo","en":""},{"ja":"Marc Moreeuw","en":""},{"ja":"Toma Hara","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Augmenting Sleep Behavior with a Wearable: Can Self-Reflection Help?"},"year":2024,"month":4,"publisher":"Augmented Humans 2024 (AHs24)","location":"Melbourne, Australia","review":"reviewed","presen_type":"poster","fields":[],"doi":"10.1145/3652920.3653049","abstract":{"ja":"Wearable devices can have potential applications to intelligence augmentation by reducing the cognitive effort of maintaining healthy habits, but they are yet to be successful at substantially modifying behavior. This paper investigates this failure through an examination of how self-reflection impacts habit transformation and\ndiscusses potential solutions for encouraging data actionability."}},{"key":"Hirabayashi2024","id":"1939","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Naoki Hirabayashi","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Zheng Cheng","en":""},{"ja":"Kazunori Minatani","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"VisPhoto: Photography for People with Visual Impairments via Omnidirectional Imaging Post-Production"},"journal":{"ja":"Proceedings of the 5th International Workshop on Digitization and E-Inclusion in Mathematics and Science 2024 (DEIMS2024)"},"num_pages":"4","year":2024,"month":2,"location":"Tokyo, Japan","review":"reviewed","presen_type":"oral","fields":["visually-impaired-assistance","object-recognition"],"abstract":{"ja":"Many people with visual impairments would like to take photographs.\nHowever, they often have difficulty pointing the camera at the target.\nIn this paper, we address this problem by proposing a novel photo-taking system called VisPhoto.\nUnlike conventional methods, VisPhoto \\textit{generates a photograph} in post-production.\nWhen the shutter button is pressed, VisPhoto captures an omnidirectional camera image that contains the surrounding scene of the camera.\nIn post-production, the system outputs a cropped region as a ``photograph'' that satisfies the user's preference."}},{"key":"Yonekura2024","id":"2009","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"米倉 崇晃","en":""},{"ja":"松本 壮史","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"岩村 雅一","en":""},{"ja":"中井 朋則","en":""},{"ja":"山内 大輔","en":""},{"ja":"唐原 一郎","en":""},{"ja":"星野 真人","en":""},{"ja":"上杉 健太朗","en":""},{"ja":"黄瀬 浩一","en":""},{"ja":"峰雪 芳宣","en":""},{"ja":"小塚 俊明","en":""}],"title":{"ja":"数理モデルを用いた、キクタニギク頭状花序パターンを変化させる要因の探究"},"journal":{"ja":"日本植物学会第88回大会予稿集"},"year":2024,"month":9,"review":"not_reviewed","fields":["plant-image-processing"]},{"key":"Yamazaki2024_CVIM","id":"1980","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"山崎 匠真","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"三輪 由佳","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"3次元位置関係を用いたブドウ果粒のマッチング"},"journal":{"ja":"情報処理学会 研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2024-CVIM-238","presen_id":"39","pages":"1-8","year":2024,"month":5,"location":"東京大学","review":"not_reviewed","presen_type":"poster","fields":["pattern-recognition","3d-reconstruction","plant-image-processing"],"url":"http://id.nii.ac.jp/1001/00234053/"},{"key":"Saimyou2024","id":"1991","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"最明 新太","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"三輪 由佳","en":""},{"ja":"岩村 雅一","en":""},{"ja":"大山克己","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"健康な葉画像のみを学習に用いたモデルに基づくブドウの葉の病変検出"},"journal":{"ja":"農業情報学会2024年度年次大会"},"presen_id":"P-03","pages":"142-143","year":2024,"month":5,"location":"高崎健康福祉大学","review":"not_reviewed","presen_type":"poster","fields":["plant-image-processing"]},{"key":"Meii2024","id":"1965","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"明井 綜佑","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"岩村 雅一","en":""},{"ja":"塚谷裕一","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"CT画像を用いたアリ植物の空洞の3次元可視化","en":"CT image-based 3D Visualization of ant plant cavities"},"journal":{"ja":"情報処理学会第86回全国大会予稿集"},"presen_id":"4T-04","num_pages":"2","year":2024,"month":3,"publisher":"情報処理学会","address":"東京都港区芝公園 3-5-8 機械振興会館 308-3","location":"神奈川大学 横浜キャンパス","review":"not_reviewed","presen_type":"oral","fields":["plant-image-processing"],"abstract":{"ja":"アリ植物は，植物体内に空洞をもち，その中にアリを住まわせている．このような特異な関係を持つ生物は互いに影響を与えながら進化した（共進化）と考えられ，これらの植物の形態情報の解析は，共進化の過程を理解する上で一つの重要な示唆を提供する．そこで本研究は，アリ植物の空洞の3次元形状を解析することを目的として，アリ植物のCT 画像から，空洞の3次元可視化をする．CT 画像にはアリ植物の領域のみが撮影されるため，3次元可視化にあたり空洞部分のみを取得する処理が必要となる．そこで本研究では，輪郭検出や凸包検出，モルフォロジー処理を組み合わせることでアリ植物の空洞を検出し，3次元可視化を実現した．"}},{"key":"Matsumoto2024_PRMU","id":"1964","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"松本壮史","en":""},{"ja":"内海ゆづ子","en":""},{"ja":"小塚俊明","en":""},{"ja":"岩村雅一","en":""},{"ja":"黄瀬浩一","en":""},{"ja":"中井朋則","en":""},{"ja":"山内大輔","en":""},{"ja":"唐原一郎","en":""},{"ja":"峰雪芳宣","en":""},{"ja":"星野真人","en":""},{"ja":"上杉健太郎","en":""}],"title":{"ja":"CT画像を用いたキクタニギクの小花配列の推定"},"journal":{"ja":"電子情報通信学会 技術研究報告"},"volume":"vol.123","number":"no.409","presen_id":"PRMU2023-75","pages":"pp.140-145","num_pages":"6","year":2024,"month":3,"review":"not_reviewed","fields":["deep-learning","plant-image-processing","object-recognition"],"url":"https://ken.ieice.org/ken/paper/202403044cb7/"},{"key":"Naka2024_IEICEPRMU","id":"1957","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"仲 勇樹","en":"Yuki Naka"},{"ja":"内海 ゆづ子","en":"Yuzuko Utsumi"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"塚谷 裕一","en":"Hirokazu Tsukaya"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"分割統治戦略に基づく花のCT画像の花弁セグメンテーション","en":"Divide-and-Conquer Strategy Based Petal Segmentation in CT Images"},"journal":{"ja":"電子情報通信学会 技術研究報告","en":"IEICE Technical Report"},"volume":"123","number":"409","presen_id":"PRMU2023-58","pages":"41-46","num_pages":"6","year":2024,"month":3,"location":"広島大学 東広島キャンパス","review":"not_reviewed","presen_type":"oral_and_poster","fields":["plant-image-processing"],"url":"https://ken.ieice.org/ken/paper/20240303ZcbH/"},{"key":"Yamada2024","id":"1961","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"山田 大記","en":""},{"ja":"井上 勝文","en":""},{"ja":"Partha Prathim Roy","en":""},{"ja":"岩村 雅一","en":""},{"ja":"吉岡 理文","en":""}],"title":{"ja":"未知手話単語獲得に向けた手話動作特徴量空間の検討"},"journal":{"ja":"電子情報通信学会 技術研究報告","en":"IEICE Technical Report"},"volume":"123","number":"409","presen_id":"PRMU2023-52","pages":"7-12","num_pages":"6","year":2024,"month":3,"location":"広島大学 東広島キャンパス","review":"not_reviewed","presen_type":"oral_and_poster","fields":["sign-language-recognition"],"url":"https://ken.ieice.org/ken/paper/20240303Zcbs/","abstract":{"ja":"近年，様々な手法が提案されている手話認識研究において大きな課題となるのが，データセット内の手話動画の語彙の不足である．この課題に対し，特に未学習の手話単語(未知単語)のデータに対して自動で学習動画の収集とラベル付けする手法が求められている．幸いYoutube等に多数字幕情報のついた手話動画が存在するため，これらの内，特定の未知単語が含まれる複数の動画より類似する動作を抽出することで，未知単語を自動獲得できるのではないかと期待されている．これを実現するために必要となりえる要素技術は，(1)類似の手話動作を精確に表現できる動作特徴量と，(2)その特徴量に基づいて，手話動画から目的の単語の部分を精確に切り出す処理である．本研究では，これの実現の第一歩として，手話単語の細かな動作を一つ一つ記述可能な3DCNNであるI3Dを用いて手話単語を表現する特徴量について検討する．本稿では，この特徴量表現の良し悪しを評価するテストヘッドとして，既知単語と未知単語を判別する問題を考え，特徴量の識別性能について評価した結果について述べる．"}},{"key":"Minatani2024","id":"1973","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"南谷 和範","en":""},{"ja":"渡辺 哲也","en":""},{"ja":"岩村雅一","en":""}],"title":{"ja":"模型の触察を行うシンポジウムのオンライン開催の試み"},"journal":{"ja":"電子情報通信学会、サイバーワールド研究会"},"year":2024,"month":3,"review":"not_reviewed","presen_type":"oral","fields":["visually-impaired-assistance"]},{"key":"Buoy2024_PhD","id":"2037","type":"doctoral_thesis","language":"EN","show":"show_except_datafile","authors":[{"ja":"Rina Buoy","en":""}],"title":{"ja":"Innovative Techniques for Reduced-Complexity Text Recognition"},"year":2024,"month":8,"review":"not_reviewed","fields":["document-image-analysis","deep-learning"]},{"key":"Utsumi2024","id":"1962","type":"survey","language":"JA","show":"show","authors":[{"ja":"内海 ゆづ子","en":""}],"title":{"ja":"5分で分かる!? 有名論文ナナメ読み：Alan Mathison Turing : The Chemical Basis of Morphogenesis"},"journal":{"ja":"情報処理"},"volume":"65","number":"3","pages":"172-175","year":2024,"month":2,"review":"not_reviewed","fields":["plant-image-processing"],"doi":"info:doi/10.20729/00232351","url":"https://ipsj.ixsq.nii.ac.jp/ej/?action=pages_view_main&active_action=repository_view_main_item_detail&item_id=232465&item_no=1&page_id=13&block_id=8"},{"key":"Ou2024","id":"1960","type":"other","language":"JA","show":"show","authors":[{"ja":"王 瑩","en":""},{"ja":"手島 知昭","en":""},{"ja":"橋本 敦史","en":""},{"ja":"早志 英朗","en":""},{"ja":"柳川 由紀子","en":""},{"ja":"山口 光太","en":""},{"ja":"横田 達也","en":""},{"ja":"内海 ゆづ子","en":""}],"title":{"ja":"MIRU2023ダイバーシティに関わるアンケート調査報告"},"journal":{"ja":"情報処理学会研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2024-CVIM-237","number":"38","num_pages":"1-7","year":2024,"month":3,"location":"広島大学東広島キャンパス","review":"not_reviewed","presen_type":"oral","fields":[]},{"key":"Neigel2024_arXiv","id":"2116","type":"other","language":"EN","show":"show","authors":[{"ja":"","en":"Peter Neigel"},{"ja":"","en":"Andrew Vargo"},{"ja":"","en":"Benjamin Tag"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Identifying Periods of Cyclical Stress in University Students Using Wearables In-the-Wild"},"journal":{"en":"arXiv"},"num_pages":"19","year":2024,"month":2,"review":"not_reviewed","fields":["wellbeing"],"url":"https://doi.org/10.48550/arXiv.2402.11823"},{"key":"Selby2024_arXiv","id":"1953","type":"other","language":"EN","show":"show","authors":[{"ja":"David Selby","en":""},{"ja":"Kai Spriestersbach","en":""},{"ja":"Yuichiro Iwashita","en":""},{"ja":"Dennis Bappert","en":""},{"ja":"Archana Warrier","en":""},{"ja":"Sumantrak Mukherjee","en":""},{"ja":"Muhammad Nabeel Asim","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Sebastian Vollmer","en":""}],"title":{"ja":"Quantitative knowledge retrieval from large language models"},"journal":{"ja":"arXiv"},"book_title":"arXiv:2402.07770v1 [cs.IR]","num_pages":"21","year":2024,"month":2,"review":"not_reviewed","fields":["information-retrieval"],"url":"https://arxiv.org/abs/2402.07770","abstract":{"ja":"Large language models (LLMs) have been extensively studied for their abilities to generate convincing natural language sequences, however their utility for quantitative information retrieval is less well understood. In this paper we explore the feasibility of LLMs as a mechanism for quantitative knowledge retrieval to aid data analysis tasks such as elicitation of prior distributions for Bayesian models and imputation of missing data. We present a prompt engineering framework, treating an LLM as an interface to a latent space of scientific literature, comparing responses in different contexts and domains against more established approaches. Implications and challenges of using LLMs as 'experts' are discussed."}},{"key":"Vargo2023b","id":"1948","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Andrew Vargo","en":""},{"ja":"Kohei Yamaguchi","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"A Context-Based Multimedia Vocabulary Learning System for Mobile Users"},"journal":{"ja":"Informatics"},"volume":"11","number":"1","year":2023,"month":12,"review":"reviewed","fields":[],"doi":"10.3390/informatics11010001","url":"https://doi.org/10.3390/informatics11010001","abstract":{"ja":"Vocabulary acquisition and retention is an essential part of learning a foreign language and many learners use flashcard applications to repetitively increase vocabulary retention. However, it can be difficult for learners to remember new words and phrases without any context. In this paper, we propose a system that allows users to acquire new vocabulary with media which gives context to the words. Theoretically, this use of multimedia context should enable users to practice with interest and increased motivation, which has been shown to enhance the effects of contextual language learning. An experiment with 46 English as foreign language learners showed better retention after two weeks with the proposed system as compared to ordinary flashcards. However, the impact was not universally beneficial to all learners. An analysis of participant attributes that were gathered through surveys and questionnaires shows a link between personality and learning traits and affinity for learning with this system. This result indicates that the proposed system provides a significant advantage in vocabulary retention for some users, while other users should stay with traditional flashcard applications. The implications of this study indicate the need for the development of more personalized learning applications."}},{"key":"Buoy2023_I_Vitstr","id":"1947","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"","en":"Rina Buoy"},{"ja":"","en":"Masakazu Iwamura"},{"ja":"","en":"Sovila Srun"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"ViTSTR-Transducer: Cross-Attention-Free Vision Transformer Transducerfor Scene Text Recognition"},"journal":{"en":"Journal of Imaging"},"volume":"9","number":"12","num_pages":"17","year":2023,"month":12,"publisher":"MDPI","review":"reviewed","fields":["document-image-analysis","pattern-recognition","deep-learning"],"doi":"10.3390/jimaging9120276","abstract":{"en":"Attention-based encoder?decoder scene text recognition (STR) architectures have been proven effective in recognizing text in the real world, thanks to their ability to learn an internal language model. Nevertheless, the cross-attention operation that is used to align visual and linguistic features during decoding is computationally expensive, especially in low-resource environments. To address this bottleneck, we propose a cross-attention-free STR framework that still learns a language model. The framework we propose is ViTSTR-Transducer, which draws inspiration from ViTSTR, a vision transformer (ViT)-based method designed for STR and the recurrent neural network transducer (RNN-T) initially introduced for speech recognition. The experimental results show that our ViTSTR-Transducer models outperform the baseline attention-based models in terms of the required decoding floating point operations (FLOPs) and latency while achieving a comparable level of recognition accuracy. Compared with the baseline context-free ViTSTR models, our proposed models achieve superior recognition accuracy. Furthermore, compared with the recent state-of-the-art (SOTA) methods, our proposed models deliver competitive results."}},{"key":"Buoy2023_ACCESS","id":"1937","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"","en":"Rina Buoy"},{"ja":"","en":"Masakazu Iwamura"},{"ja":"","en":"Sovila Srun"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Towards A Low-Resource Non-Latin-Complete Baseline: An Exploration of Khmer Optical Character Recognition"},"journal":{"en":"IEEE ACCESS"},"volume":"11","num_pages":"18","year":2023,"month":11,"publisher":"IEEE","review":"reviewed","fields":["pattern-recognition","document-image-analysis","deep-learning"],"doi":"10.1109/ACCESS.2023.3332361","abstract":{"en":"Many existing text recognition methods rely on the structure of Latin characters and words. Such methods may not be able to deal with non-Latin scripts that have highly complex features, such as character stacking, diacritics, ligatures, non-uniform character widths, and writing without explicit word boundaries. In addition, from a natural language processing (NLP) perspective, most non-Latin languages are considered low-resource due to the scarcity of large-scale data. This paper presents a convolutional Transformer-based text recognition method for low-resource non-Latin scripts, which uses local two-dimensional (2D) feature maps. The proposed method can handle images of arbitrarily long textlines, which may occur with non-Latin writing without explicit word boundaries, without resizing them to a fixed size by using an improved image chunking and merging strategy. It has a low time complexity in self-attention layers and allows efficient training. The Khmer script is used as the representative of non-Latin scripts because it shares many features with other non-Latin scripts, which makes the construction of an optical character recognition (OCR) method for Khmer as hard as that for other non-Latin scripts. Thus, by analogy with the AI-complete concept, a Khmer OCR method can be considered as one of the non-Latin-complete methods and can be used as a low-resource non-Latin baseline method. The proposed 2D method was trained on synthetic datasets and outperformed the baseline models on both synthetic and real datasets. Fine-tuning experiments using Khmer handwritten palm leaf manuscripts and other non-Latin scripts demonstrated the feasibility of transfer learning from the Khmer OCR method. To contribute to the low-resource language community, the training and evaluation datasets will be made publicly available."}},{"key":"Buoy2023_I_Explainable","id":"1938","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"","en":"Rina Buoy"},{"ja":"","en":"Masakazu Iwamura"},{"ja":"","en":"Sovila Srun"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Explainable Connectionist-Temporal-Classification-Based Scene Text Recognition"},"journal":{"en":"Journal of Imaging"},"volume":"9","number":"11","num_pages":"20","year":2023,"month":11,"publisher":"MDPI","review":"reviewed","fields":["pattern-recognition","document-image-analysis","deep-learning","object-recognition"],"doi":"10.3390/jimaging9110248","abstract":{"en":"Connectionist temporal classification (CTC) is a favored decoder in scene text recognition (STR) for its simplicity and efficiency. However, most CTC-based methods utilize one-dimensional (1D) vector sequences, usually derived from a recurrent neural network (RNN) encoder. This results in the absence of explainable 2D spatial relationship between the predicted characters and corresponding image regions, essential for model explainability. On the other hand, 2D attention-based methods enhance recognition accuracy and offer character location information via cross-attention mechanisms, linking predictions to image regions. However, these methods are more computationally intensive, compared with the 1D CTC-based methods. To achieve both low latency and model explainability via character localization using a 1D CTC decoder, we propose a marginalization-based method that processes 2D feature maps and predicts a sequence of 2D joint probability distributions over the height and class dimensions. Based on the proposed method, we newly introduce an association map that aids in character localization and model prediction explanation. This map parallels the role of a cross-attention map, as seen in computationally-intensive attention-based architectures. With the proposed method, we consider a ViT-CTC STR architecture that uses a 1D CTC decoder and a pretrained vision Transformer (ViT) as a 2D feature extractor. Our ViT-CTC models were trained on synthetic data and fine-tuned on real labeled sets. These models outperform the recent state-of-the-art (SOTA) CTC-based methods on benchmarks in terms of recognition accuracy. Compared with the baseline Transformer-decoder-based models, our ViT-CTC models offer a speed boost up to 12 times regardless of the backbone, with a maximum 3.1% reduction in total word recognition accuracy. In addition, both qualitative and quantitative assessments of character locations estimated from the association map align closely with those from the cross-attention map and ground-truth character-level bounding boxes."}},{"key":"Vargo2023_IEEE","id":"1936","type":"journal_paper","language":"EN","show":"show_except_datafile","authors":[{"ja":"Andrew Vargo","en":""},{"ja":"Peter Neigel","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Considering Wearable Health Tracking Devices and Pandemic Preparedness for Universities"},"journal":{"ja":"IEEE Pervasive Computing"},"volume":"22","number":"4","pages":"35-44","year":2023,"month":10,"review":"reviewed","fields":["sensing"],"doi":"10.1109/MPRV.2023.3322460","abstract":{"ja":"In the future, it is possible that universities could use sensing devices to help manage students and classroom disruptions during pandemics. Wearable sensing devices have shown the capability to identify oncoming illnesses. In particular, the Oura Ring, a smart-ring with sleep-tracking features, can detect cases of COVID-19 before users are aware of it. While these devices show promise, they are only effective for pandemic preparedness and response if users pay attention to the outputs of the devices and consistently use them. In this article, we discuss the results of a year-long in-the-wild study with 35 participants at a university who wore Oura Rings. After an orientation, the participants wore their rings with no restrictions or minimum wearing requirements. By retroactively looking at how participants used the rings for monitoring their health, we identify strategies and potential problems with employing wearables for health monitoring in universities."}},{"key":"Baray2023","id":"2010","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"S. B. Baray","en":""},{"ja":"M. U. Ahmed","en":""},{"ja":"M. E. H. Chowdhury","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"EOG-Based Reading Detection in the Wild Using Spectrograms and Nested Classification Approach"},"journal":{"ja":"IEEE Access"},"volume":"11","pages":"105619-105632","year":2023,"month":9,"publisher":"IEEE","review":"reviewed","fields":["learning-analytics","sensing"],"doi":"10.1109/ACCESS.2023.3316032"},{"key":"Vargo2023a","id":"1932","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Andrew Vargo","en":""},{"ja":"Chris Blakely","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Poverty Traps in Online Knowledge-Based Peer-Production Communities"},"journal":{"ja":"Informatics"},"volume":"10","number":"3","year":2023,"month":7,"review":"reviewed","fields":[],"doi":"10.3390/informatics10030061","abstract":{"ja":"Online knowledge-based peer-production communities, like question and answer sites (Q&A), often rely on gamification, e.g., through reputation points, to incentivize users to contribute frequently and effectively. These gamification techniques are important for achieving the critical mass that sustains a community and enticing new users to join. However, aging communities tend to build “poverty traps” that act as barriers for new users. In this paper, we present our investigation of 32 domain communities from Stack Exchange and our analysis of how different subjects impact the development of early user advantage. Our results raise important questions about the accessibility of knowledge-based peer-production communities. We consider the analysis results in the context of changing information needs and the relevance of Q&A in the future. Our findings inform policy design for building more equitable knowledge-based peer-production communities and increasing the accessibility to existing ones."}},{"key":"Nolasco2023","id":"1935","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Hannah R Nolasco","en":""},{"ja":"Andrew Vargo","en":""},{"ja":"Niklas Bohley","en":""},{"ja":"Christian Brinkhaus","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Examining Participant Adherence with Wearables in an In-the-Wild Setting"},"journal":{"ja":"Sensors"},"volume":"23","number":"14","year":2023,"month":7,"review":"reviewed","fields":[],"doi":"10.3390/s23146479","abstract":{"ja":"Wearable devices offer a wealth of data for ubiquitous computing researchers. For instance, sleep data from a wearable could be used to identify an individual’s harmful habits. Recently, devices which are unobtrusive in size, setup, and maintenance are becoming commercially available. However, most data validation for these devices come from brief, short-term laboratory studies or experiments which have unrepresentative samples that are also inaccessible to most researchers. For wearables research conducted in-the-wild, the prospect of running a study has the risk of financial costs and failure. Thus, when researchers conduct in-the-wild studies, the majority of participants tend to be university students. In this paper, we present a month-long in-the-wild study with 31 Japanese adults who wore a sleep tracking device called the Oura ring. The high device usage results found in this study can be used to inform the design and deployment of longer-term mid-size in-the-wild studies."}},{"key":"Mahad2023_IEICE","id":"1901","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Fairuz Safwan Mahad","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Learning multi-level features for improved 3D reconstruction"},"journal":{"ja":"IEICE Transactions on Information and Systems"},"volume":"E106-D","number":"3","pages":"381-390","num_pages":"9","year":2023,"month":3,"publisher":"IEICE","review":"reviewed","fields":["3d-reconstruction"],"doi":"10.1587/transinf.2020ZDP7508","abstract":{"ja":"3D reconstruction methods using neural networks are popular and have been studied extensively. However, the resulting models typically lack detail, reducing the quality of the 3D reconstruction. This is because the network is not designed to capture the fine details of the object. Therefore, in this paper, we propose two networks designed to capture both the coarse and fine details of the object to improve the reconstruction of the detailed parts of the object. To accomplish this, we design two networks. The first network uses a multi-scale architecture with skip connections to associate and merge features from other levels. For the second network, we design a multi-branch deep generative network that separately learns the local features, generic features, and the intermediate features through three different tailored components. In both network architectures, the principle entails allowing the network to learn features at different levels that can reconstruct the fine parts and the overall shape of the reconstructed 3D model. We show that both of our methods outperformed state-of-the-art approaches."}},{"key":"Kinose2023","id":"1902","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Rikuya Kinose","en":""},{"ja":"Yuzuko Utsumi","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Tiller estimation method using deep neural networks"},"journal":{"ja":"Frontiers in Plant Science"},"volume":"13:1016507","num_pages":"11","year":2023,"month":1,"review":"reviewed","fields":["plant-image-processing"],"doi":"10.3389/fpls.2022.1016507"},{"key":"Oshiba2023","id":"1903","type":"journal_paper","language":"EN","show":"show_except_datafile","authors":[{"ja":"","en":"Junki Oshiba"},{"ja":"","en":"Motoi Iwata"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Face Image Generation of Anime Characters Using an Advanced First Order Motion Model with Facial Landmarks"},"journal":{"en":"IEICE Transactions on Information and Systems"},"volume":"E106-D","number":"1","pages":"22-30","num_pages":"9","year":2023,"month":1,"review":"reviewed","fields":["comic-computing"],"doi":"10.1587/transinf.2022MUP0004","url":"https://search.ieice.org/bin/summary.php?id=e106-d_1_22&category=D&year=2023&lang=E&abst=","abstract":{"en":"Recently, deep learning for image generation with a guide for the generation has been progressing. Many methods have been proposed to generate the animation of facial expression change from a single face image by transferring some facial expression information to the face image. In particular, the method of using facial landmarks as facial expression information can generate a variety of facial expressions. However, most methods do not focus on anime characters but humans. Moreover, we attempted to apply several existing methods to anime characters by training the methods on an anime character face dataset; however, they generated images with noise, even in regions where there was no change. The first order motion model (FOMM) is an image generation method that takes two images as input and transfers one facial expression or pose to the other. By explicitly calculating the difference between the two images based on optical flow, FOMM can generate images with low noise in the unchanged regions. In the following, we focus on the aspect of the face image generation in FOMM. When we think about the employment of facial landmarks as targets, the performance of FOMM is not enough because FOMM cannot use a facial landmark as a facial expression target because the appearances of a face image and a facial landmark are quite different. Therefore, we propose an advanced FOMM method to use facial landmarks as a facial expression target. In the proposed method, we change the input data and data flow to use facial landmarks. Additionally, to generate face images with expressions that follow the target landmarks more closely, we introduce the landmark estimation loss, which is computed by comparing the landmark detected from the generated image with the target landmark. Our experiments on an anime character face image dataset demonstrated that our method is effective for landmark-guided face image generation for anime characters. Furthermore, our method outperformed other methods quantitatively and generated face images with less noise."}},{"key":"Hirabayashi2023","id":"1931","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Naoki Hirabayashi","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Zheng Cheng","en":""},{"ja":"Kazunori Minatani","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"VisPhoto: Photography for People with Visual Impairments via Post-Production of Omnidirectional Camera Imaging"},"journal":{"ja":"Proceedings of the 25th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS)"},"pages":"1-17","year":2023,"month":10,"location":"New York, USA","review":"reviewed","presen_type":"oral","fields":["visually-impaired-assistance","object-recognition"],"doi":"10.1145/3597638.3608422","awards":{"en":["SIGACCESS Best Paper Award Winner"]},"abstract":{"ja":"Many people with visual impairments would like to take photographs. However, they often have difficulty pointing the camera at the target. In this paper, we address this problem by proposing a novel photo-taking system called VisPhoto. Unlike conventional methods, VisPhoto generates a photograph in post-production. When the shutter button is pressed, VisPhoto captures an omnidirectional camera image that contains the surrounding scene of the camera. In post-production, the system outputs a cropped region as a “photograph” that satisfies the user’s preference. We conducted an experiment consisting of two parts. First, 24 people with visual impairments took photographs with a genuine iPhone camera app, a conventional method, and VisPhoto. Second, 20 sighted people evaluated the quality of the photographs. The experimental results showed that the participants with visual impairments preferred to use VisPhoto to take photographs of difficult targets, whereas they preferred the conventional method for easy targets. Moreover, we revealed that their preferences for photo-taking methods were influenced by the participants’ needs and values about photography and their confidence in their photographic abilities."}},{"key":"Neigel2023","id":"1934","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Peter Neigel","en":""},{"ja":"Andrew Vargo","en":""},{"ja":"Yusuke Komatsu","en":""},{"ja":"Chris Blakely","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Exploring Users’ Ability to Choose a Proper Fit in Smart-Rings: A Year-Long “In the Wild” Study"},"journal":{"ja":"Interact 2023 (IFIP Conference on Human-Computer Interaction)"},"year":2023,"month":8,"review":"reviewed","presen_type":"poster","fields":[],"doi":"10.1007/978-3-031-42293-5_59","abstract":{"ja":"Wearables have seen an increase in popularity as the technology has advanced. Via physiological sensing, these devices allow individuals to easily monitor their daily health, but there are limitations in real-world implementation. One limitation is the ability of end users to properly gauge how well a device fits. Proper fit is often crucial for accurate data collection, as ill-fitting devices can result in poor performance. In this work, we examine the fit of the ring sizes the participants chose at the start of the in-the-wild study and examine the effect it has on sensor data. An analysis of 38 participants shows that only every third participant chooses an optimal ring size. We show that this leads to reduced efficacy of the wearable."}},{"key":"Nolasco2023_IFIP","id":"1933","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Hannah R. Nolasco","en":""},{"ja":"Andrew Vargo","en":""},{"ja":"Yusuke Komatsu","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Perception Versus Reality: How User Self-reflections Compare to Actual Data"},"journal":{"ja":"Interact 2023 (IFIP Conference on Human-Computer Interaction)"},"year":2023,"month":8,"review":"reviewed","presen_type":"oral","fields":[],"doi":"10.1007/978-3-031-42286-7_37","abstract":{"ja":"One of the main promises of wearable sensing devices capable of physiological tracking is the potential that users can leverage the technology to make positive life changes. Now that these devices have sufficient accuracy, it is feasible that users could make decisions based on feedback from the device to change their habits and improve their well-being. A potential challenge to this, however, is whether users are able to recognize actual changes in their behavior compared to perceived changes. In this work, we look at participants who used the Oura ring as part of an in-the-wild study. Based on data analysis following 10?12 months of usage, we find that users who had positive perceptions of habit change towards sleep failed to show long-term improvement in their sleep quality, indicating a gap between perceived success and real-world impact. This work is important to informing the future design and implementation of health-tracking wearables."}},{"key":"Kotoura2023","id":"1942","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"琴浦 陽南","en":"Hina Kotoura"},{"ja":"山本 航平","en":"Kohei Yamamoto"},{"ja":"堀川 昴輝","en":"Koki Horikawa"},{"ja":"高本 凌平","en":"Ryohei Takamoto"},{"ja":"野々宮 悠太","en":"Yuta Nonomiya"},{"ja":"岩下 雄一郎","en":"Yuichiro Iwashita"},{"ja":"栗生 薫","en":"Kaori Kuriu"},{"ja":"石丸 翔也","en":"Shoya Ishimaru"},{"ja":"仲子 聡一郎","en":"Soichiro Nakako"},{"ja":"岡村 浩史","en":"Hiroshi Okamura"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"新谷 歩","en":"Ayumi Shintani"}],"title":{"ja":"臨床テーブルデータと胸部X線画像を統合したツリー系機械学習モデルによるCOVID-19患者の予後予測","en":"Tree-based machine learning model integrating clinical table data and chest x-ray images to predict prognosis of covid-19 patients"},"journal":{"ja":"第43回医療情報学連合大会 (第24回日本医療情報学会学術大会) 論文集","en":"Proceedings of the 43rd Joint Conference on Medical Informatics (the 24th conference of the Japan Association for Medical Informatics)"},"presen_id":"4-G-4-03","num_pages":"5","year":2023,"month":11,"publisher":"日本医療情報学会","location":"神戸ファッションマート","review":"reviewed","presen_type":"oral","fields":[],"url":"https://confit.atlas.jp/guide/event/jcmi2023/subject/4-G-4-03/entries","abstract":{"en":"Background: Since the peak number of COVID-19 mortality in Japan tends to increase with each outbreak, it is important to identify patients at high risk of death at an early stage. In a previous study, a neural network model was simultaneously trained with clinical and imaging information to predict the prognosis of COVID-19 patients. On the other hand, it has been reported that tree-based models outperform neural networks for table data. We aimed to construct and evaluate a tree-based prognostic model that treats chest X-ray images in the same way as other clinical information by reducing them to one dimension using neural networks. \nMethods: This study used the Stony Brook University COVID-19 dataset of 1313 inpatients. In order to predict mortality of all patients, we developed a model (hereafter referred to as “proposed model”) with LightGBM trained with the clinical table data combined with the result of prediction of a model with DenseNet trained with chest radio graphs. We also developed a model with LightGBM trained with the clinical data alone and a model with DenseNet trained with images. \nResults: The f1 scores for the proposed model, the table data-only model, and the image-only model were 0.59, 0.58, and 0.25, respectively. The ROC-AUC for the proposed model, the table data-only model, and the image-only model were 0.91, 0.93, and 0.63, respectively. Image prediction had the highest variable importance in the proposed model. \nConclusion: the proposed model achieved high prediction accuracy. It was suggested that clinical table data contributed more to than chest X-ray images. On the other hand, images may improve prediction accuracy when combined with clinical table data."}},{"key":"Liu2023","id":"1940","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"劉 少楠","en":"Shaonan Liu"},{"ja":"野々宮 悠太","en":"Yuta Nonomiya"},{"ja":"岩下 雄一郎","en":"Yuichiro Iwashita"},{"ja":"木村 誠一","en":"Seiichi Kimura"},{"ja":"齋藤 美紅","en":"Miku Saito"},{"ja":"山崎 桜花","en":"Oka Yamasaki"},{"ja":"石丸 翔也","en":"Shoya Ishimaru"},{"ja":"仲子 聡一郎","en":"Soichiro Nakako"},{"ja":"岡村 浩史","en":"Hiroshi Okamura"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"新谷 歩","en":""}],"title":{"ja":"集中治療室における再入室イベントを予測する機械学習ベースの生存時間解析モデルに時系列データが与える影響","en":"Effects of time-series data on machine learning-based survival analysis models for predicting readmission events in intensive care units"},"journal":{"ja":"第43回医療情報学連合大会 (第24回日本医療情報学会学術大会) 論文集","en":"Proceedings of the 43rd Joint Conference on Medical Informatics (the 24th conference of the Japan Association for Medical Informatics)"},"presen_id":"4-G-4-05","num_pages":"5","year":2023,"month":11,"publisher":"日本医療情報学会","location":"神戸ファッションマート","review":"reviewed","presen_type":"oral","fields":[],"url":"https://confit.atlas.jp/guide/event/jcmi2023/subject/4-G-4-05/entries","abstract":{"en":"One of the problems in the intensive care unit (ICU) is the lack of ICU beds. In proper bed management, it is important to predict readmission events, in which a patient who has been discharged is readmitted to the ICU. There have been many studies using survival analysis models for predicting readmission events in an ICU, which consider the last examination results such as vital signs and blood tests. However, there are few reports using survival analysis models for predicting readmission events in an ICU, which consider the past examination results as time-series features. On the other hand, clinicians treat these time-series data as important prognostic factors. Since we think it is meaningful to consider the impact of using time-series data as features in survival analysis models, we used Random Survival Forest (RSF) and DeepSurv as survival analysis models to build a prediction model for ICU readmissions for ICU discharge patients and investigated the impact of time-series data on the model. As a result, we show that time-series data can improve the prediction accuracy of survival time analysis models and survival analysis models considering time-series data may assist clinicians in their predictions and contribute to appropriate bed management."}},{"key":"Nonomiya2023","id":"1941","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"野々宮 悠太","en":"Yuta Nonomiya"},{"ja":"柳 健大","en":"Kenta Yanagi"},{"ja":"栗生 薫","en":"Kaori Kuriu"},{"ja":"内田 時央","en":"Tokio Uchida"},{"ja":"岩下 雄一郎","en":"Yuichiro Iwashita"},{"ja":"藤原 史規","en":"Fuminori Fujiwara"},{"ja":"山岡 幹太","en":"Kanta Yamaoka"},{"ja":"石丸 翔也","en":"Shoya Ishimaru"},{"ja":"仲子 聡一郎","en":"Soichiro Nakako"},{"ja":"岡村 浩史","en":"Hiroshi Okamura"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"新谷 歩","en":"Ayumi Shintani"}],"title":{"ja":"集中治療室における短期的な死亡を予測する再帰型ニューラルネットワークを用いた動的予測モデルの提案：MIMIC-IVデータベースを利用した後方視的コホート研究","en":"Recurrent neural network-based dynamic prediction model for short-term mortality in intensive care units: retrospective cohort study using the MIMIC-IV database"},"journal":{"ja":"第43回医療情報学連合大会 (第24回日本医療情報学会学術大会) 論文集","en":"In Proceedings of the 43rd Joint Conference on Medical Informatics (the 24th conference of the Japan Association for Medical Informatics)"},"presen_id":"4-G-4-06","num_pages":"5","year":2023,"month":11,"publisher":"日本医療情報学会","location":"神戸ファッションマート","review":"reviewed","fields":[],"url":"https://confit.atlas.jp/guide/event/jcmi2023/subject/4-G-4-06/entries","abstract":{"en":"Objectives: This study aims to develop a dynamic prediction model for short-term mortality based on static and time-series information of patients in the ICU.\nMethods: This retrospective study used the MIMIC-IV database and included critically ill patients between 19 and 89 years old with at least a 48-hour ICU stay. They were randomly assigned to train, validation, and test in proportions of 70%, 15%, and 15%, respectively. Age, gender, race, vital signs, and time-series data, including blood laboratory values, were extracted as characteristics. Missing values were imputed with the last observations. RNN-based models were trained to predict 24-hour mortality from each time point. Predictive performance was evaluated using the area under the ROC curve. \nResults: 21637 patients were included, 9264 (42.8%) were female, the median age was 66 (55-76) years, the median length of stay was 90 (63-157) hours, and 1946 (9.0%) patients died within 24 hours from ICU discharge, The area under the ROC curve in the test was 0.882, 0.885, and 0.891 for simple RNN, LSTM, and GRU models, respectively. \nConclusions: We developed an RNN-based dynamic prediction model that accurately predicts short-term mortality in the ICU. We need further evaluation for the impact on patient outcomes and decision-making processes to implement the model in the real world."}},{"key":"Iwashita2023_IEICE","id":"1906","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"岩下 雄一郎","en":"Yuichiro Iwashita"},{"ja":"岩田 基","en":"Motoi Iwata"},{"ja":"Andrew W. Vargo","en":"Andrew W. Vargo"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"英単語学習を長期的に継続するための適応型学習システムの開発","en":"Developing an Adaptive Learning System for Long-term English Vocabulary Learning"},"journal":{"ja":"2023年電子情報通信学会総合大会講演論文集","en":"Proceedings of the 2023 IEICE General Conference"},"presen_id":"D-15-10","num_pages":"1","year":2023,"month":3,"publisher":"電子情報通信学会","location":"芝浦工業大学大宮キャンパス","review":"not_reviewed","presen_type":"oral","fields":["learning-analytics","sensing"],"url":"https://www.ieice.org/publications/conferences/summary.php?id=CONF0000139324&expandable=0&ConfCd=2023G&session_num=D-15&lecture_number=D-15-10&year=2023&conf_type=G","abstract":{"ja":"英単語の学習は英語習得の基本であり，長期的に継続して学習することが重要である．長期的に学習を継続するには学習の動機づけが必要であり，興味のある教材を用いることでこれが可能だと期待できる．しかし，興味のある教材を用いた英単語学習システムは開発されているが，学習に時間がかかってしまうため，学習効率が下がってしまう課題がある．本稿では，学習者の学習中のパフォーマンスや振る舞いに基づいて，学習者に最適な教材を推薦することで，長期的に効率的な学習を継続可能なシステムを開発したので報告する．"}},{"key":"Tagawa2023_mtg","id":"1922","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"田川 直弥","en":"Naoya Tagawa"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"南谷 和範","en":"Kazunori Minatani"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"視覚障害者を対象としたディスプレイ操作を必要とする家庭用3Dプリンタ操作支援システム"},"journal":{"ja":"ヒューマンインタフェース学会研究報告集"},"volume":"25","number":"2","presen_id":"mtg-ap-0243","pages":"31-40","num_pages":"10","year":2023,"month":3,"publisher":"NPO法人ヒューマンインタフェース学会","location":"京都工芸繊維大学","review":"not_reviewed","presen_type":"oral_and_poster","fields":["visually-impaired-assistance"],"abstract":{"ja":"近年登場した家庭用3Dプリンタは、手で物の形状を理解する触察をより身近にする可能性を秘めている．しかし、印刷に必要なディスプレイの操作は視覚に大きく依存しているため、視覚障害者が3Dプリンタを扱うのは難しい．そこで3Dプリンタのディスプレイの表示を認識、音声出力する操作支援システムを提案する．予備実験では3Dプリンタの操作タスクを，スマートフォンの認識アプリと提案システムをそれぞれ用いて行ってもらい，操作完了までにかかる時間の比較を行った．その結果，提案システムの方が操作完了までにかかる時間が短くなった．"}},{"key":"Ohara2023","id":"1924","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"大原健太郎","en":"Kentaro Ohara"},{"ja":"Andrew Vargo","en":"Andrew Vargo"},{"ja":"黄瀬浩一","en":"Koichi Kise"}],"title":{"ja":"個人に適した有酸素運動の単語記憶課題への効果"},"journal":{"ja":"情報処理学会全国大会公演論文集"},"num_pages":"2","year":2023,"month":3,"review":"not_reviewed","presen_type":"oral","fields":["learning-analytics"]},{"key":"Nakajima2023_CLE","id":"1908","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"中島 悠太","en":""},{"ja":"岩田 基","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"text-to-imageモデルによる生成画像を用いた英単語学習の提案と画像で表現可能な英単語の調査"},"journal":{"ja":"情報処理学会研究報告教育学習支援情報システム（CLE)"},"volume":"2023-CLE-39","pages":"1-6","year":2023,"month":3,"review":"not_reviewed","presen_type":"oral","fields":[]},{"key":"Uragami2023_NC","id":"1921","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"浦上 太一","en":"Taichi Uragami"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"バーチャルペット飼育型単語学習システムの提案"},"journal":{"ja":"情報処理学会第85回全国大会"},"year":2023,"month":3,"review":"not_reviewed","fields":["learning-analytics"]},{"key":"Uragami2023","id":"1920","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"浦上 太一","en":"Taichi Uragami"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"バーチャルペット飼育型単語学習システムとその評価"},"journal":{"ja":"第202回ヒューマンコンピュータインタラクション研究発表会"},"year":2023,"month":3,"review":"not_reviewed","presen_type":"oral","fields":["learning-analytics"]},{"key":"Nakatani2023","id":"1913","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"中谷友哉","en":""},{"ja":"内海ゆづ子","en":""},{"ja":"藤本仰一","en":""},{"ja":"岩村雅一","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"花画像から花弁配置を推定するWebアプリケーションの構築"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"122","number":"404","presen_id":"PRMU2022-105","pages":"242-247","year":2023,"month":3,"location":"はこだて未来大学","review":"not_reviewed","presen_type":"oral_and_poster","fields":["plant-image-processing"]},{"key":"Iwamura2023","id":"1974","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村雅一","en":""},{"ja":"河合隆哲","en":""},{"ja":"高嶋慶伍","en":""},{"ja":"南谷和範","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"受動的な情報取得における適切な情報要約方法の検討"},"journal":{"ja":"情報アクセシビリティをめぐる諸問題に関する研究集会"},"year":2023,"month":2,"review":"not_reviewed","presen_type":"oral","fields":["visually-impaired-assistance"]},{"key":"Mahad2023_PhD","id":"1930","type":"doctoral_thesis","language":"EN","show":"show_except_abstract","authors":[{"ja":"Fairuz Safwan MAHAD","en":""}],"title":{"ja":"Engineering Features for 3D Reconstruction"},"year":2023,"month":2,"review":"not_reviewed","fields":["3d-reconstruction"]},{"key":"Tagawa2023_BThesis","id":"1923","type":"bachelors_thesis","language":"JA","show":"show","authors":[{"ja":"田川直弥","en":"Naoya Tagawa"}],"title":{"ja":"視覚障害者を対象とした3Dプリンタ操作支援システム"},"num_pages":"43","year":2023,"month":2,"review":"not_reviewed","presen_type":"oral","fields":["visually-impaired-assistance"]},{"key":"Iwamura2023_J","id":"1946","type":"survey","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":""}],"title":{"ja":"研究の「そのとき」を考える"},"journal":{"ja":"電子情報通信学会誌"},"volume":"107","number":"12","pages":"1074-1077","year":2023,"month":12,"publisher":"電子情報通信学会","review":"not_reviewed","fields":[],"url":"https://www.journal.ieice.org/summary.php?id=k106_12_1074","abstract":{"ja":"研究を続けていると，多くの研究者に影響を与える「当たり」の研究に出会うことがある．そのような研究は，どうして当たったのであろうか？　ともすれば研究内容だけに注目してしまいがちであるが，その理由の一端はそれを生み出した時代背景にあるのかもしれない．本小特集では，過去の「当たり」の研究がどのように生まれ，その後の研究シーンにどのような影響を与えたのかを，ゆかりのある研究者に解説して頂く．本稿では，その企画意図と「当たり」の研究テーマの生まれ方の一例を述べ，更に研究活動のライフサイクルという考え方を紹介する．"}},{"key":"Ookura2023","id":"1912","type":"survey","language":"JA","show":"show","authors":[{"ja":"大倉 史生","en":""},{"ja":"郭 威","en":""},{"ja":"戸田 陽介","en":""},{"ja":"内海 ゆづ子","en":""}],"title":{"ja":"イマドキノ植物とCV CV の新たな地平を拓くのは･･･植物!?"},"book_title":"コンピュータビジョン最前線 Spring 2023","pages":"9-47","year":2023,"month":3,"publisher":"共立出版株式会社","review":"not_reviewed","fields":["plant-image-processing"]},{"key":"Utsumi2023_FiT","id":"2062","type":"other","language":"JA","show":"show","authors":[{"ja":"内海 ゆづ子","en":""}],"title":{"ja":"「農学×情報学」情報学の立場から"},"journal":{"ja":"第22回情報科学技術フォーラム（FiT 2023）"},"year":2023,"month":9,"review":"not_reviewed","presen_type":"oral","fields":["plant-image-processing"],"url":"https://www.ipsj.or.jp/event/fit/fit2023/abstract/data/html/event/event_C-1.html","note":{"ja":"招待講演"},"abstract":{"ja":"第22回情報科学技術フォーラム（FiT 2023）イベント企画際を究める 〜「×情報学」による分野横断型研究@OMU 講演・パネルディスカッション"}},{"key":"Utsumi2023_Symp","id":"1911","type":"other","language":"JA","show":"show","authors":[{"ja":"内海 ゆづ子","en":""}],"title":{"ja":"生食ブドウ栽培支援のための画像認識技術","en":"Image recognition techniques for suporting table grape farmers"},"journal":{"ja":"第10回計測自動制御学会制御部門マルチシンポジウム予稿集"},"presen_id":"2M3-6","num_pages":"5","year":2023,"month":3,"location":"立命館大学びわこくさつキャンパス","review":"not_reviewed","presen_type":"oral","fields":["plant-image-processing"],"note":{"ja":"招待講演"}},{"key":"Muta2022","id":"1874","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Kaoru Muta","en":""},{"ja":"Shiho Takata","en":""},{"ja":"Yuzuko Utsumi","en":""},{"ja":"Atsushi Matsumura","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"TAIM: Tool for Analyzing Root Images to Calculate the Infection Rate of Arbuscular Mycorrhizal Fungi"},"journal":{"ja":"Frontiers in Plant Science"},"volume":"13","num_pages":"11","year":2022,"month":5,"review":"reviewed","fields":["plant-image-processing"],"doi":"10.3389/fpls.2022.881382"},{"key":"Vargo2022_IEEE","id":"1852","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Andrew Vargo","en":""},{"ja":"Shoya Ishimaru","en":""},{"ja":"Md. Rabiul Islam","en":""},{"ja":"Benjamin Tag","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Obtaining Labels for In-the-Wild Studies:Using Visual Cues and Recall"},"journal":{"ja":"IEEE Pervasive"},"volume":"21","number":"1","pages":"9-17","year":2022,"month":1,"review":"reviewed","fields":["learning-analytics"],"doi":"10.1109/MPRV.2021.3129500","abstract":{"ja":"The observer effect found in laboratory studies has long posed a problem for\nresearchers. In-the-wild studies reduce the observer effect, but have problems with gathering accurately labeled data usable for training algorithms. Manual labeling is time-consuming, obtrusive, and unfeasible, and if done by the researchers, it potentially violates the privacy of the participants. In this article, we present a labeling workflow based on an in-the-wild study that investigated cognitive state changes through eye-gaze in naturalistic settings. We contribute a setup that enables participants to label their data unobtrusively and quickly. We use J!NS MEME electrooculography glasses, Narrative Clip 2 wearable cameras, and a proprietary data tagging software package. Our setup is reproducible for field studies, preserves data integrity, and maintains participant privacy. This workflow can be extended to other studies in pervasive and ubiquitous computing and is especially suitable for deployment in the pandemic and postpandemic world."}},{"key":"Islam2022_IEICE","id":"1892","type":"letter","language":"EN","show":"show_except_datafile","authors":[{"ja":"Md. Rabiul Islam","en":""},{"ja":"Andrew W. Vargo","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Exploring Sensor Modalities to Capture User Behaviors for Reading Detection"},"journal":{"ja":"IEICE Transactions on Information and Systems"},"volume":"E105-D","number":"9","pages":"1629-1633","num_pages":"5","year":2022,"month":9,"publisher":"IEICE","review":"reviewed","fields":["learning-analytics","deep-learning","activity-recognition"],"doi":"10.1587/transinf.2020ZDL0003"},{"key":"Mahad2022","id":"1850","type":"letter","language":"EN","show":"show","authors":[{"ja":"Fairuz Safwan Mahad","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Learning Pyramidal Feature Hierarchy for 3D reconstruction"},"journal":{"ja":"IEICE Transactions on Information and Systems"},"volume":"E105-D","number":"2","pages":"446-449","num_pages":"4","year":2022,"month":2,"publisher":"IEICE","review":"reviewed","fields":["3d-reconstruction"],"doi":"10.1587/transinf.2020ZDL0001","abstract":{"ja":"Neural network-based three-dimensional (3D) reconstruction methods have produced promising results.\nHowever, they do not pay particular attention to reconstructing detailed parts of objects.\nThis occurs because the network is not designed to capture the fine details of objects.\nIn this paper, we propose a network designed to capture both the coarse and fine details of objects to improve the reconstruction of the fine parts of objects."}},{"key":"Vargo2022_ISWCa","id":"1889","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Andrew W. Vargo","en":""},{"ja":"Masayuki Fujimoto","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Sleep Quality Tracking and Alignment with a Simple In-the-Wild Task"},"journal":{"ja":"Adjunct Proceedings of the 2022 ACM International Joint Conference on Pervasive and Ubiquitous Computing and the 2022 International Symposium on Wearable Computers (UbiComp/ISWC '22 Adjunct)"},"num_pages":"3","year":2022,"month":9,"publisher":"ACM","location":"Cambridge, UK","review":"reviewed","presen_type":"poster","fields":["learning-analytics","sensing"],"abstract":{"ja":"The introduction of wearable sleep trackers, such as the Oura Ring, allow users to monitor the duration and quality of their sleep. For researchers in the fields of education and cognitive sciences, these wearables promise a new source of valuable data, especially if the wearables can be used outside of the laboratory. A problem in validating the wearable data for daily life is that it is difficult to give tasks which are both effective enough challenging users, but simple and quick enough that they can be completed by participants daily. In this poster, we present and in-the-wild study which employs a simple arithmetic task. The results indicate that sleep quality data from a wearable sleep trackers can be used to gauge performance."}},{"key":"Vargo2022_ISWCb","id":"1890","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Andrew W. Vargo","en":""},{"ja":"Kentaro Ohara","en":""},{"ja":"Takumi Nakai","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Just the Right Amount of Exercise: Aerobic Exercise and Memory"},"journal":{"ja":"Adjunct Proceedings of the 2022 ACM International Joint Conference on Pervasive and Ubiquitous Computing and the 2022 International Symposium on Wearable Computers (UbiComp/ISWC '22 Adjunct)"},"num_pages":"3","year":2022,"month":9,"publisher":"ACM","location":"Cambridge, UK","review":"reviewed","presen_type":"poster","fields":["learning-analytics","sensing"],"abstract":{"ja":"A popular perception is that aerobic exercises are effective for improving performance on memory tasks. However, previous studies have not yet explored the effectiveness of aerobic exercise for  individuals. We investigate the effects of aerobic exercise on English vocabulary learning by managing the conditions under which aerobic exercise is effective for individuals. We use a heart-rate sensor to control the amount of exertion a participant made on a treadmill while learning the vocabulary words. The results show this method can improve memory retention. This indicates that close monitoring of individual aerobic exertion through physiological sensing can make improved learning with exercise a reality."}},{"key":"Vargo2022_ISWCc","id":"2011","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Andrew Vargo","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Mathilde Hutin","en":""},{"ja":"Sofia Kobylyanskaya","en":""},{"ja":"Ioana Vasilescu","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"K. Watanabe","en":""},{"ja":"Shoya Ishimaru","en":""},{"ja":"Benjamin Tag","en":""},{"ja":"Tilman Dingler","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Laurence Devillers","en":""},{"ja":"Andreas Dengel","en":""}],"title":{"ja":"Learning Cyclotron: An Ecosystem of Knowledge Circulation"},"journal":{"ja":"Adjunct Proceedings of the 2022 ACM International Joint Conference on Pervasive and Ubiquitous Computing and the 2022 International Symposium on Wearable Computers (UbiComp/ISWC '22 Adjunct)"},"pages":"308-312","year":2022,"month":9,"publisher":"ACM","review":"reviewed","fields":["learning-analytics"]},{"key":"Vargo2022_ISWCd","id":"2012","type":"intl_conf","language":"JA","show":"show","authors":[{"ja":"Andrew Vargo","en":""},{"ja":"V. Abou-Khalil","en":""},{"ja":"Shoya Ishimaru","en":""},{"ja":"Benjamin Tag","en":""},{"ja":"Mathilde Hutin","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"Laurence Devillers","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Delivering Sensing Technologies for Education and Learning"},"journal":{"ja":"Adjunct Proceedings of the 2022 ACM International Joint Conference on Pervasive and Ubiquitous Computing and the 2022 International Symposium on Wearable Computers (UbiComp/ISWC '22 Adjunct)"},"pages":"263-266","year":2022,"month":9,"publisher":"ACM","review":"reviewed","fields":["learning-analytics","sensing"]},{"key":"Yamaoka2022","id":"1888","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Kanta Yamaoka","en":""},{"ja":"Andrew W. Vargo","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Towards Guided-Breath as an Intervention Within Computer-Assisted Learning"},"journal":{"ja":"Adjunct Proceedings of the 2022 ACM International Joint Conference on Pervasive and Ubiquitous Computing and the 2022 International Symposium on Wearable Computers (UbiComp/ISWC '22 Adjunct)"},"num_pages":"3","year":2022,"month":9,"publisher":"ACM","location":"Cambridge, UK","review":"reviewed","presen_type":"poster","fields":["learning-analytics","sensing"],"abstract":{"ja":"Mindfulness has been actively investigated in HCI. While its positive effects on our cognitive functions are reported, carrying out mindfulness activities in general  requires self-discipline to get its reported benefits. Our study aims to integrate a visual guide of breathing into computer-assisted learning to help the learner achieve better performance. We present within-subject research with 10 participants who learned sets of Croatian vocabulary with a multiple-choice interface showing either for 10 seconds : (a) guided-breath or (b) break instruction after every question.   Our hypothesis was that (a) brings a better learner's score than (b). However, statistical analysis of learners' quiz scores did not support our hypothesis, which brought us two lessons for future improvements of our work-in-progress research. Such intervention (i) could lead to distraction rather than concentration without careful quantity and quality, and (ii) might need personalization depending on individual mindfulness levels."}},{"key":"Suzawa2022","id":"1925","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Haruki Suzawa","en":""},{"ja":"Ko Watanabe","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"Shoya Ishimaru","en":""}],"title":{"ja":"Supporting Smooth Interruption in a Video Conference by Dynamically Changing Background Music Depending on the Amount of Utterance"},"journal":{"ja":"Proceedings of the 2022 ACM International Joint Conference on Pervasive and Ubiquitous Computing"},"pages":"299-302","year":2022,"month":9,"location":"Atlanta, USA","review":"reviewed","fields":["activity-recognition"],"doi":"10.1145/3544793.3560384"},{"key":"Islam2022_IIAIAAI","id":"1893","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Md. Rabiul Islam","en":""},{"ja":"Andrew W. Vargo","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Evaluating Contrastive Learning for Fine-grained Reading Detection"},"journal":{"ja":"Proc. 12th International Congress on Advanced Applied Informatics (IIAI-AAI)"},"pages":"430-435","year":2022,"month":7,"location":"Kanazawa City","review":"reviewed","fields":["activity-recognition","deep-learning"],"doi":"10.1109/IIAIAAI55812.2022.00091"},{"key":"Watanabe2022","id":"1887","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Tetsuya Watanabe","en":""},{"ja":"Kazunori Minatani","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Effect of Visual and/or Haptic Experience on Haptic 3D Model Recognition"},"journal":{"ja":"Proc. ICCHP-AAATE 2022 Open Access Compendium \"Assistive Technology, Accessibility and (e)Inclusion\", Part 1"},"pages":"42-47","year":2022,"month":7,"location":"Lecco, Italy","review":"reviewed","presen_type":"oral","fields":["visually-impaired-assistance"],"doi":"10.35011/icchp-aaate22-p1-06"},{"key":"Iwamura2022_ICCHPAAATE","id":"1855","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Takaaki Kawai","en":""},{"ja":"Keigo Takashima","en":""},{"ja":"Kazunori Minatani","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Acquiring Surrounding Visual Information without Actively Taking Photos for People with Visual Impairment"},"journal":{"ja":"Proc. Joint International Conference on Digital Inclusion, Assistive Technology & Accessibility (ICCHP-AAATE 2022)"},"volume":"13341","series":"Lecture Notes in Computer Science","pages":"229-240","year":2022,"month":7,"location":"Lecco, Italy","review":"reviewed","presen_type":"oral","fields":["visually-impaired-assistance","object-recognition"],"doi":"10.1007/978-3-031-08648-9_27","abstract":{"ja":"Recent advancements in recognition technology allow people with visual impairment to obtain visual information from their surroundings using smartphone apps and assistive devices. This paper points out a problem with this approach that has not attracted much attention. That is, the user is required to actively take a photo, which is not always easy for people with visual impairment. To address this problem, in contrast to the current standard approach, which we call active information acquisition, we propose passive information acquisition (PIA), which does not require the user to actively take a photo. However, PIA creates a new problem. That is, the app tends to transfer too much information to the user. Therefore, this paper explores better ways for people with visual impairment toward obtaining only the desired visual information in PIA. Specifically, we experimented with nine people with visual impairment to evaluate seven information transmission methods, including information summarization and interactive communication methods."}},{"key":"Higashimura2022_CHI","id":"1854","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Riku Higashimura","en":""},{"ja":"Andrew W. Vargo","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Helping Mobile Learners Know Unknown Words through their Reading Behavior"},"journal":{"ja":"CHI Conference on Human Factors in Computing Systems Extended Abstracts (CHI '22 Extended Abstracts)"},"num_pages":"5","year":2022,"month":4,"location":"New Orleans, LA, USA","review":"reviewed","presen_type":"poster","fields":["learning-analytics"],"doi":"10.1145/3491101.3519620","url":"https://dl.acm.org/doi/abs/10.1145/3491101.3519620","abstract":{"ja":"Vocabulary acquisition is a fundamental part of learning a new language. In order to acquire new vocabulary, words with meanings that are unknown to the learner must be added to the language learning process. When searching for material in the target language, it is useful to know how much of a document is made up of currently unknown words. One simple way to estimate the unknown words in a document is to use the frequency of occurrence, which indicates the difficulty of the word. However, this approach can lead to missed unknown words. In this study, we aim to improve the accuracy of unknown word estimation by using reading activity data obtained from smartphone sensors and taking into account the individual learner's English reading behavior. We apply a novel user interface which allows us to improve estimation through reading behavior, without the use of eye-trackers."}},{"key":"Higashimura2022_FIT","id":"1898","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"東村 理功","en":"Riku Higashimura"},{"ja":"Andrew W. Vargo","en":"Andrew W. Vargo"},{"ja":"岩田 基","en":"Motoi Iwata"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"モバイル学習者の読書行動に基づく未知単語推定","en":"Unknown Word Estimation Based on Mobile Learners’ Reading Behavior"},"journal":{"ja":"第21回情報科学技術フォーラム(FIT2022)","en":"FIT2022"},"presen_id":"CK-002","num_pages":"4","year":2022,"month":9,"location":"慶応義塾大学 矢上キャンパス","review":"not_reviewed","presen_type":"oral","fields":["learning-analytics"],"abstract":{"ja":"語彙の習得は，新しい言語を学習する際の基本的な要素である．スマートフォンの普及により，語学学習者は日々のスケジュールの中でより柔軟に学習できるようになっている．新たな語彙を習得するためには，学習者個人が意味を知らない単語(以下，未知単語) を何らかの方法で語学学習の対象にしなくてはならない．文書中の未知単語を推定する簡単な方法として，その単語の難易度を示す出現頻度を用いる方法がある．しかし，この方法では未知単語の見落としが発生する可能性がある．本研究では，スマートフォンのセンサーから取得した読書行動データを用い，学習者個人の読書行動を考慮することで，未知単語推定の精度を向上させることを目指す．","en":"Vocabulary acquisition is a fundamental component of learning a new language. The widespread use of smartphones allows language learners more flexibility in their daily schedules. In order to acquire new vocabulary, words whose meanings are unknown to the individual learner (hereafter referred to as \"unknown words\") must be included in the language learning process in some way. A simple method for estimating unknown words in a document is to use the frequency of occurrence, which indicates the difficulty level of the word. However, this method may result in missing unknown words. In this study, we aim to improve the accuracy of unknown word estimation by using reading behavior data obtained from smartphone sensors and taking into account the reading behavior of individual learners."}},{"key":"Minatani2022","id":"1975","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"南谷和範","en":""},{"ja":"渡辺哲也","en":""},{"ja":"岩村雅一","en":""}],"title":{"ja":"立体教材を用いた遠隔ワークショップの可能性と課題 - 触察オンラインシンポジウムの経験から -"},"journal":{"ja":"LIFE2022"},"year":2022,"month":8,"review":"not_reviewed","presen_type":"oral","fields":["visually-impaired-assistance"]},{"key":"Shiroi2022_Symp","id":"1896","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"城居 巧","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"継続学習における宝くじ仮説に基づく破滅的忘却の抑制"},"journal":{"ja":"第25回画像の認識・理解シンポジウム予稿集"},"volume":"IS1-34","year":2022,"month":7,"location":"兵庫県姫路市","review":"not_reviewed","presen_type":"poster","fields":["deep-learning"]},{"key":"Takamoto2022","id":"1897","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"高本 凌平","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"大山 克己","en":""},{"ja":"岩村 雅一","en":""},{"ja":"北宅 善昭","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"宇宙空間における植物生育観察のための局所特徴量を用いた対応領域検出"},"journal":{"ja":"第25回画像の認識・理解シンポジウム予稿集"},"volume":"IS2-61","year":2022,"month":7,"location":"兵庫県姫路市","review":"not_reviewed","presen_type":"poster","fields":["plant-image-processing","deep-learning"]},{"key":"Tamura2022_IS","id":"1895","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"Yasuto Tamura","en":""},{"ja":"Yuzuko Utsumi","en":""},{"ja":"Yuka Miwa","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Unsupervised Monocular Depth Estimation with Omnidirectional Camera"},"journal":{"ja":"第25回画像の認識・理解シンポジウム予稿集","en":"Proceedings of the 25th Meeting on Image Recognition and Understanding"},"presen_id":"IS1-15","year":2022,"month":7,"location":"兵庫県姫路市","review":"not_reviewed","presen_type":"poster","fields":["plant-image-processing","deep-learning"]},{"key":"Matsumoto2022","id":"1881","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"松本 壮史","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"小塚 俊明","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""},{"ja":"中井 朋則","en":""},{"ja":"山内 大輔","en":""},{"ja":"唐原 一郎","en":""},{"ja":"峰雪 芳宣","en":""},{"ja":"星野 真人","en":""},{"ja":"上杉 健太郎","en":""}],"title":{"ja":"キクの花の構造推定のためのCT画像セグメンテーション"},"book_title":"情報処理学会研究報告コンピュータビジョンとイメージメディア（CVIM）","volume":"2022-CVIM-230","number":"17","pages":"1-7","num_pages":"7","year":2022,"month":5,"location":"豊田工業大学 (ハイブリッド開催)","review":"not_reviewed","presen_type":"poster","fields":["plant-image-processing","deep-learning"],"url":"http://id.nii.ac.jp/1001/00217722/"},{"key":"Yamada2022","id":"1949","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"山田 大記","en":""},{"ja":"井上 勝文","en":""},{"ja":"岩村 雅一","en":""},{"ja":"Partha Pratim Roy","en":""},{"ja":"吉岡 理文","en":""}],"title":{"ja":"時空間特徴量を用いた手話単語認識における未知単語判定手法の検討"},"journal":{"ja":"情報処理学会 研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2022-CVIM-230","number":"14","pages":"1-8","year":2022,"month":5,"location":"豊田工業大学 (ハイブリッド開催)","review":"not_reviewed","presen_type":"poster","fields":["sign-language-recognition"],"url":"http://id.nii.ac.jp/1001/00217719/","abstract":{"ja":"手話単語認識にはラベル付き学習用データが必要となる．しかし，手話動画のラベル付けには専門知識が必要なため，データには限りがある．そのため，自動で手話単語動画をラベル付けする手法が求められるが，これを実現するには高精度な認識モデルが必要であるというジレンマがある．学習済み単語 (既知単語) のデータ数増加に対しては，既存モデルを活用できるかもしれないが，学習外単語 (未知単語)に対しては，既知単語のいずれかに認識しようとする既存モデルの活用は困難である．この問題に対応するためには，既知単語か未知単語かで処理が異なるため，まずラベルの付いていない手話単語動画の手話動作が既知単語か未知単語かを区別する必要がある．本研究では，未知単語動画の有効活用のために，ラベルの付いていない手話単語動画が未知単語かを判定する手法を提案する．具体的には，3DCNN ベースの手法により手話動作を時空間特徴量列で表し，特徴量同士の類似度を求めることで，未知単語かを判定する．"}},{"key":"Hirano2022","id":"1880","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"平野 甫","en":"Hajime Hirano"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"学習済みモデルに任意の出力を促す画像生成とそれを用いた学習"},"journal":{"ja":"情報処理学会研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2022-CVIM-230","number":"24","pages":"1-6","num_pages":"6","year":2022,"month":5,"location":"豊田工業大学（ハイブリッド開催）","review":"not_reviewed","presen_type":"poster","fields":["deep-learning"],"url":"http://id.nii.ac.jp/1001/00217729/"},{"key":"Naka2022_CVIM","id":"1878","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"仲 勇樹","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"岩村 雅一","en":""},{"ja":"塚谷 裕一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"花の3次元形状推定のためのCT画像を用いた花弁セグメンテーション"},"journal":{"ja":"情報処理学会研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2022-CVIM-230","number":"13","pages":"1-8","num_pages":"8","year":2022,"month":5,"location":"豊田工業大学（ハイブリッド開催）","review":"not_reviewed","fields":["plant-image-processing"],"url":"http://id.nii.ac.jp/1001/00217718/"},{"key":"Shiroi2022_CVIM","id":"1879","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"城居巧","en":"Takumi Shiroi"},{"ja":"岩村雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬浩一","en":"Koichi Kise"}],"title":{"ja":"宝くじ仮説を用いた継続学習における破滅的忘却の抑制","en":"Lottery Ticket Hypothesis for Overcoming Forgetting on continualLearning"},"journal":{"ja":"情報処理学会研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2022-CVIM-230","number":"28","pages":"1-8","num_pages":"8","year":2022,"month":5,"location":"豊田工業大学（ハイブリッド開催）","review":"not_reviewed","fields":["deep-learning"],"url":"http://id.nii.ac.jp/1001/00217733/"},{"key":"Tamura2022","id":"1858","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"田村 泰斗","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"三輪 由佳","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"ブドウ房の３次元復元のための全方位カメラを使用した教師なし単眼深度推定"},"book_title":"情報処理学会研究報告コンピュータビジョンとイメージメディア（CVIM）","volume":"2022-CVIM-229(5)","pages":"1-8","year":2022,"month":3,"review":"not_reviewed","fields":[]},{"key":"Riku2022_IEICE","id":"1853","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"東村 理功","en":"Higashimura Riku"},{"ja":"岩田 基","en":"Iwata Motoi"},{"ja":"黄瀬 浩一","en":"Kise Koichi"}],"title":{"ja":"スマートフォンを用いた読書活動データからの未知単語推定","en":"Estimating Unknown Words from Reading Activity Data Using Smartphones"},"journal":{"ja":"2022年 電子情報通信学会 総合大会 講演論文集","en":"IEICE Conferences Archives"},"presen_id":"D-15-10","num_pages":"1","year":2022,"month":3,"location":"オンライン","review":"not_reviewed","presen_type":"oral","fields":["learning-analytics"],"abstract":{"ja":"語彙の習得は語学学習の基本である．\nスマートフォンの普及により，語学学習者は日々のスケジュールの中でより柔軟に学習できるようになっている．\n新たな語彙を習得するためには，学習者個人が意味を知らない単語(以下，未知単語)を何らかの方法で語学学習の対象にしなくてはならない．\nそこで，文書中の未知単語を推定する指標の一つとして，単語の難しさを表す出現頻度を用いる方法がある．\n一般に，単語の出現頻度が低い単語ほど，その単語が利用される機会が少ないため，どの学習者にとっても未知単語である可能性が高い．しかし，出現頻度は全ての学習者に共通する客観的な指標である．\nそのため，出現頻度の高い簡単な単語が未知単語である場合，学習の必要性が高いにも関わらず，出現頻度のみに基づいた学習では学習漏れの可能性がある．\nしたがって，一人一人の学習者に応じた未知単語を推定する必要がある．\n学習者固有の読書活動データの一つに読書時間がある．読書時間は，学習者の英文読解能力の指標として有効であることが報告されている．\n\n本研究では学習者に依存しない文書特徴量に，スマートフォンのセンサデータで得られる読書活動データを組み込むことで未知単語推定の精度向上を目指す．\n実験の結果，実験参加者14名のうち12名に対して，読書活動データが未知単語推定に寄与することを確認した．","en":"Vocabulary acquisition is a fundamental part of learning a new language. In order to acquire new vocabulary, words with meanings that are unknown to the learner must be added to the language learning process. When searching for material in the target language, it is useful to know how much of a document is made up of currently unknown words. One simple way to estimate the unknown words in a document is to use the frequency of occurrence, which indicates the difficulty of the word. However, this approach can lead to missed unknown words. In this study, we aim to improve the accuracy of unknown word estimation by using reading activity data obtained from smartphone sensors and taking into account the individual learner's English reading behavior. We apply a novel user interface which allows us to improve estimation through reading behavior, without the use of eye-trackers."}},{"key":"Tanaka2022","id":"1872","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"田中 勢也","en":""},{"ja":"Andrew W. Vargo","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"腕の動きとWebカメラを用いた英語問題の確信度推定"},"journal":{"ja":"2022年電子情報通信学会総合大会 講演論文集"},"presen_id":"D-15-12","year":2022,"month":3,"review":"not_reviewed","presen_type":"oral","fields":["learning-analytics"]},{"key":"Yuasa2022_CLE","id":"1860","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"湯浅 成章","en":""},{"ja":"Andrew Vargo","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"個人に適した英語多肢選択問題の自動生成方法の提案"},"journal":{"ja":"情報処理学会研究報告教育学習支援情報システム（CLE)"},"volume":"2022-CLE-36","number":"5","pages":"1-8","year":2022,"month":3,"review":"not_reviewed","fields":["learning-analytics"]},{"key":"Nakai2022_CLE","id":"1861","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"中井 拓海","en":""},{"ja":"Andrew Vargo","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"習慣添付型マイクロラーニングの提案"},"journal":{"ja":"情報処理学会研究報告教育学習支援情報システム（CLE）"},"volume":"2022-CLE-36","number":"7","num_pages":"1-8","year":2022,"month":3,"review":"not_reviewed","fields":["learning-analytics"]},{"key":"Yamaguchi2022_CLE","id":"1859","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"山口 航平","en":""},{"ja":"岩田 基","en":""},{"ja":"Andrew Vargo","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"マルチメディアVocabulometerによる動画の字幕を利用した英単語学習"},"journal":{"ja":"情報処理学会研究報告教育学習支援情報システム（CLE）"},"volume":"2022-CLE-36","number":"4","pages":"1-8","year":2022,"month":3,"review":"not_reviewed","fields":[]},{"key":"Islam2022_PhD","id":"1943","type":"doctoral_thesis","language":"EN","show":"show_except_abstract","authors":[{"ja":"Md. Rabiul Islam","en":""}],"title":{"ja":"In-The-Wild Reading Activity Evaluation UsingSelf-Supervision"},"year":2022,"month":7,"review":"not_reviewed","fields":["activity-recognition"]},{"key":"Iwamura2022","id":"1954","type":"survey","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":""},{"ja":"平林 直樹","en":""},{"ja":"程 征","en":""},{"ja":"南谷 和範","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"おめでとう論文賞：全方位カメラでの撮影とその後処理による視覚障害者の写真撮影支援"},"journal":{"ja":"電子情報通信学会 情報・システムソサイエティ誌"},"volume":"27","number":"3","pages":"11","year":2022,"month":11,"review":"not_reviewed","fields":[],"url":"https://www.ieice.org/iss/iss_r/jpn/wp/wp-content/themes/iss/assets/pdf/society_mag/Vol27No3.pdf"},{"key":"Utsumi2022","id":"1900","type":"survey","language":"JA","show":"show","authors":[{"ja":"内海ゆづ子","en":"Yuzuko Utsumi"}],"title":{"ja":"画像で植物を測る"},"journal":{"ja":"計測と制御"},"volume":"61","number":"10","pages":"717-718","year":2022,"month":10,"review":"not_reviewed","fields":["plant-image-processing"]},{"key":"Ambai2022","id":"1848","type":"survey","language":"JA","show":"show","authors":[{"ja":"安倍 満","en":""},{"ja":"岩村 雅一","en":""}],"title":{"ja":"パターン認識・メディア理解の研究最前線"},"journal":{"ja":"情報・システムソサイエティ誌"},"volume":"105","year":2022,"month":2,"publisher":"電子情報通信学会","review":"not_reviewed","fields":[]},{"key":"Utsumi2022_Confc","id":"1910","type":"other","language":"JA","show":"show","authors":[{"ja":"内海 ゆづ子","en":""}],"title":{"ja":"画像認識技術を用いた植物形質計測"},"journal":{"ja":"第39回日本植物バイオテクノロジー学会大会・シンポジウム予稿集"},"year":2022,"month":9,"location":"堺（大阪公立大学中百舌鳥キャンパス）","review":"not_reviewed","presen_type":"oral","fields":["plant-image-processing"],"note":{"ja":"招待講演"}},{"key":"Kise2022","id":"1877","type":"other","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""}],"title":{"ja":"Reading of \"Reading\" for \"Actuating\": Augmenting Human Learning by \"Experiential Supplements\""},"journal":{"ja":"Workshop on Words/Machines #4"},"year":2022,"month":3,"review":"not_reviewed","fields":["activity-recognition","learning-analytics"],"note":{"ja":"Keynote Talk"}},{"key":"Utsumi2022_Confb","id":"1928","type":"other","language":"JA","show":"show","authors":[{"ja":"内海 ゆづ子","en":""}],"title":{"ja":"画像におけるパターン認識技術を用いた植物器官の形状データの計測"},"journal":{"ja":"第69回日本生態学会大会 シンポジウム S27 植物フェノタイピングは何を明らかにするか"},"year":2022,"month":3,"review":"not_reviewed","presen_type":"oral","fields":["plant-image-processing"],"note":{"ja":"招待講演"}},{"key":"Chakladar2021_IEEE","id":"1846","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Debashis Das Chakladar","en":""},{"ja":"Partha Pratim Roy","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"EEG-Based Cognitive State Classification and Analysis of Brain Dynamics Using Deep Ensemble Model and Graphical Brain Network"},"journal":{"ja":"IEEE Transactions on Cognitive and Developmental Systems"},"volume":"14","number":"4","pages":"1507-1519","year":2021,"month":9,"review":"reviewed","fields":["pattern-recognition"],"doi":"10.1109/TCDS.2021.3116079","abstract":{"ja":"Estimating the cognitive state of an operator has drawn increasing attention in recent years. Cognitive behavior has been effectively studied by analyzing electroencephalogram (EEG) signals. EEG has an excellent temporal resolution but poor spatial resolution; therefore, it cannot be efficiently used for the cognitive state assessment. To obtain the spatial as well as the temporal resolution of the EEG signal, we propose to combine the Filter Bank Common Spatial Pattern (FBCSP) method and Long Short-Term Memory (LSTM)-based deep ensemble model for classifying the cognitive state of a user during the mental arithmetic experiment. The proposed deep ensemble model produces 87.04% classification accuracy, and it achieves an improvement of 2.04% over the state-of-the-art method. In addition, we propose an algorithm that enables us to identify the brain dynamics for each cognitive state, which is equally important to the estimation of the cognitive state in cognitive neuroscience. For each cognitive state, the proposed information flow algorithm constructs a graphical brain network using functional and effective brain connectivity patterns, which indicates the connection strength as well as the causal effects between two anatomically separated brain regions. The proposed information flow algorithm provides an effective way of communication between different brain regions."}},{"key":"Islam2021_IMWUT","id":"1844","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Md. Rabiul Islam","en":""},{"ja":"Shuji Sakamoto","en":""},{"ja":"Yoshihiro Yamada","en":""},{"ja":"Andrew W. Vargo","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Self-supervised Learning for Reading Activity Classification"},"journal":{"ja":"Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (ACM IMWUT journal)"},"volume":"5","number":"3","presen_id":"105","pages":"1-22","year":2021,"month":9,"publisher":"ACM","review":"reviewed","fields":["activity-recognition"],"doi":"10.1145/3478088","abstract":{"ja":"Reading analysis can relay information about user's confidence and habits and can be used to construct useful feedback. A lack of labeled data inhibits the effective application of fully-supervised Deep Learning (DL) for automatic reading analysis. We propose a Self-supervised Learning (SSL) method for reading analysis. Previously, SSL has been effective in physical human activity recognition (HAR) tasks, but it has not been applied to cognitive HAR tasks like reading. We first evaluate the proposed method on a four-class classification task on reading detection using electrooculography datasets, followed by an evaluation of a two-class classification task of confidence estimation on multiple-choice questions using eye-tracking datasets. Fully-supervised DL and support vector machines (SVMs) are used as comparisons for the proposed SSL method. The results show that the proposed SSL method is superior to the fully-supervised DL and SVM for both tasks, especially when training data is scarce. This result indicates the proposed method is the superior choice for reading analysis tasks. These results are important for informing the design of automatic reading analysis platforms."}},{"key":"Iwamura2021_IEICE_Photography","id":"1835","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"平林 直樹","en":"Naoki Hirabayashi"},{"ja":"程 征","en":"Zheng Cheng"},{"ja":"南谷 和範","en":"Kazunori Minatani"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"全方位カメラでの撮影とその後処理による視覚障害者の写真撮影支援","en":"Photography for People with Visual Impairment by Photo-Taking with Omni-Directional Camera and Its Post-Production"},"journal":{"ja":"電子情報通信学会論文誌D","en":"Trans. IEICE"},"volume":"J104-D","number":"8","pages":"663-677","year":2021,"month":8,"review":"reviewed","fields":["visually-impaired-assistance"],"doi":"10.14923/transinfj.2020JDP7069","awards":{"ja":["第78回（令和3年度） 電子情報通信学会 論文賞受賞"],"en":["2021 IEICE Best Paper Award Winner"]},"note":{"ja":"論文のPDFファイルを<a href=\"http://www.ieice.org/jpn/trans_online/\">電子情報通信学会 Transactions online</a>からダウンロードすることができます．","en":"PDF file is available at <a href=\"http://www.ieice.org/jpn/trans_online/\">IEICE Transactions online (in Japanese)</a>."},"abstract":{"ja":"本論文では，全盲の視覚障害者が独力で写真を撮れる全く新しいシステムVisPhotoを提案する．スマートフォンとソーシャル・ネットワーキング・サービス（SNS）の普及により，晴眼者（目が見える人）は写真撮影と共有が手軽にできるようになった．実は視覚障害者も同じように写真を介したコミュニケーションを望んでいるが，独力での写真撮影は容易でない．特に，（全盲の）視覚障害者にとって，カメラを被写体に向けるのは難しい．この問題の解決を目指して，写真撮影に用いるスマートフォンの向きを音声や振動で利用者に指示する写真撮影支援システムが提案されている．それに対して，本論文で提案するシステムは，全方位カメラがカメラの周囲360度の景色を撮影できることを利用して，利用者の周囲を全方位画像として撮影する．そして，その後処理として，利用者の希望を反映し，かつ見た目が良い領域を「写真」として切り取る．提案手法のシステムの有用性と操作性を8名の全盲の視覚障害者に，撮影した「写真」の質を10名の晴眼者にそれぞれ検証してもらい，いずれも肯定的な評価を得た．","en":"In this paper, we introduce a novel system, which is named VisPhoto, that assists people with visual impairment in solely photographing. Though it is known that people with visual impairment want to take photos, they are not good at or even unable to include the subject in the camera frame. Existing photo-taking assistance systems give voice or sound instruction to the user so that the user can take a good photograph. The proposed system utilizes an omni-directional camera that can take a photograph containing the surrounding scene of the camera. As its post-processing, a region which reflects the user's preference and maximizes the beauty score is cropped as a ``photo.'' The proposed method obtained positive reputation through two user studies; (1) the effectiveness and user interface of the proposed system were evaluated by eight people with visual impairment; (2) the quality of ``photos'' generated by the proposed system was evaluated by ten sighted people."}},{"key":"Iwamura2021_IEICE","id":"1820","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Shunsuke Mori","en":""},{"ja":"Koichiro Nakamura","en":""},{"ja":"Takuya Tanoue","en":""},{"ja":"Yuzuko Utsumi","en":""},{"ja":"Yasushi Makihara","en":""},{"ja":"Daigo Muramatsu","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Yasushi Yagi","en":""}],"title":{"ja":"Individuality-preserving Silhouette Extraction for Gait Recognition and Its Speedup"},"journal":{"ja":"IEICE Transactions on Information and Systems"},"volume":"E104-D","number":"7","pages":"992-1001","year":2021,"month":7,"review":"reviewed","fields":["pattern-recognition"],"doi":"10.1587/transinf.2020ZDP7500","abstract":{"ja":"Most gait recognition approaches rely on silhouette-based representations due to high recognition accuracy and computational efficiency. A fundamental problem for those approaches is how to  extract individuality-preserved silhouettes from real scenes accurately. Foreground colors may be similar to background colors, and the background is cluttered. Therefore, we propose a method of individuality-preserving silhouette extraction for gait recognition using standard gait models (SGMs) composed of clean silhouette sequences of various training subjects as shape priors. The SGMs are smoothly introduced into a well-established graph-cut segmentation framework. Experiments showed that the proposed method achieved better silhouette extraction accuracy by more than 2.3% than representative methods and better identification rate of gait recognition (improved by more than 11.0% at rank 20). Besides, to reduce the computation cost, we introduced approximation in the calculation of dynamic programming. As a result, without reducing the segmentation accuracy, we reduced 85.0% of the computational cost."}},{"key":"Chakladar2021","id":"1836","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Debashis Das Chakladar","en":""},{"ja":"Pradeep Kumar","en":""},{"ja":"Shubham Mandal","en":""},{"ja":"Partha Pratim Roy","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Byung-Gyu Kim","en":""}],"title":{"ja":"3D Avatar Approach for Continuous Sign Movement Using Speech/Text"},"journal":{"ja":"Applied Sciences (Special Issue on Artificial Intelligence for Multimedia Signal Processing)"},"volume":"11","number":"8","presen_id":"3439","num_pages":"13","year":2021,"month":4,"publisher":"MDPI","review":"reviewed","fields":["pattern-recognition"],"doi":"10.3390/app11083439","abstract":{"ja":"Sign language is a visual language for communication used by hearing-impaired people with the help of hand and finger movements. Indian Sign Language (ISL) is a well-developed and standard way of communication for hearing-impaired people living in India. However, other people who use spoken language always face difficulty while communicating with a hearing-impaired person due to lack of sign language knowledge. In this study, we have developed a 3D avatar-based sign language learning system that converts the input speech/text into corresponding sign movements for ISL. The system consists of three modules. Initially, the input speech is converted into an English sentence. Then, that English sentence is converted into the corresponding ISL sentence using the Natural Language Processing (NLP) technique. Finally, the motion of the 3D avatar is defined based on the ISL sentence. The translation module achieves a 10.50 SER (Sign Error Rate) score."}},{"key":"Imagawa2021","id":"1762","type":"journal_paper","language":"EN","show":"show_except_datafile","authors":[{"ja":"","en":"Hiroyuki IMAGAWA"},{"ja":"","en":"Motoi IWATA"},{"ja":"","en":"Koichi KISE"}],"title":{"en":"Digital Watermarking Method for Printed Matters Using Deep Learning for Detecting Watermarked Areas"},"journal":{"en":"IEICE Transactions on Information and Systems"},"volume":"E104-D","number":"1","pages":"34-42","num_pages":"9","year":2021,"month":1,"review":"reviewed","fields":["digital-watermarking","deep-learning"],"doi":"10.1587/transinf.2020MUP0004","url":"https://www.jstage.jst.go.jp/article/transinf/E104.D/1/E104.D_2020MUP0004/_article"},{"key":"Tanaka2021","id":"1839","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Seiya Tanaka","en":"Seiya Tanaka"},{"ja":"Andrew W. Vargo","en":"Andrew W. Vargo"},{"ja":"Motoi Iwata","en":"Motoi Iwata"},{"ja":"Koichi Kise","en":"Koichi Kise"}],"title":{"ja":"Confidence Estimation via Wrist Movement","en":"Confidence Estimation via Wrist Movement"},"journal":{"ja":"Proceedings of the 2021 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of the 2021 ACM International Symposium on Wearable Computers (UbiComp/ISWC '21 Adjunct)","en":"Proceedings of the 2021 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of the 2021 ACM International Symposium on Wearable Computers (UbiComp/ISWC '21 Adjunct)"},"num_pages":"2","year":2021,"month":9,"review":"reviewed","presen_type":"poster","fields":["learning-analytics","sensing"],"doi":"10.1145/3460418.3479286"},{"key":"Kato2021","id":"1817","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Jin Kato","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Manga Vocabulometer, A new support system for extensive reading with Japanese mangatranslated into English"},"journal":{"ja":"MANPU2020"},"book_title":"Pattern Recognition. ICPR International Workshops and Challenges","volume":"VI","series":"LNCS","pages":"223-235","num_pages":"13","year":2021,"month":1,"publisher":"Springer Nature","address":"7th Floor | Block - C | Hardy Tower | Ramanujan IT City | Rajiv Gandhi Salai (OMR) | Taramani | Chennai - 600 113 | India |","location":"オンライン","review":"reviewed","fields":["comic-computing","sensing"]},{"key":"Oshiba2021","id":"1818","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"","en":"Junki Oshiba"},{"ja":"","en":"Motoi Iwata"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Automatic Landmark-guided Face Image Generation for Anime Characters Using C2GAN"},"journal":{"ja":"MANPU2020"},"book_title":"Pattern Recognition. ICPR International Workshops and Challenges","volume":"VI","series":"Lecture Notes in Computer Science (LNCS)","pages":"236-249","num_pages":"14","year":2021,"month":1,"publisher":"Springer Nature","address":"7th Floor | Block - C | Hardy Tower | Ramanujan IT City | Rajiv Gandhi Salai (OMR) | Taramani | Chennai - 600 113 | India |","location":"オンライン","review":"reviewed","presen_type":"oral","fields":["comic-computing","deep-learning"]},{"key":"Akai2021","id":"1757","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Ryota Akai","en":""},{"ja":"Yuzuko Utsumi","en":""},{"ja":"Yuka Miwa","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"en":"Distortion-Adaptive Grape Bunch Counting for Omnidirectional Images"},"journal":{"ja":"Proc. of the 25th International Conference on Pattern Recognition (ICPR2020)"},"pages":"599-606","year":2021,"month":1,"location":"Online","review":"reviewed","presen_type":"poster","fields":["plant-image-processing","pattern-recognition"],"doi":"10.1109/ICPR48806.2021.9412659"},{"key":"Chakladar2021_ICPR","id":"1789","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Debashis Das Chakladar","en":""},{"ja":"Shubhashis Dey","en":""},{"ja":"Partha Pratim Roy","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"EEG-Based Cognitive State Assessment Using Deep Ensemble Model and Filter Bank Common Spatial Pattern"},"journal":{"ja":"Proc. of the 25th International Conference on Pattern Recognition (ICPR2020)"},"pages":"4107-4114","year":2021,"month":1,"location":"Online","review":"reviewed","fields":["pattern-recognition"],"doi":"10.1109/ICPR48806.2021.9412869"},{"key":"Kumar2021","id":"1790","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Puneet Kumar","en":""},{"ja":"Sidharth Jain","en":""},{"ja":"Balasubramanian Raman","en":""},{"ja":"Partha Pratim Roy","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"End-to-end Triplet Loss based Emotion Embedding System for Speech Emotion Recognition"},"journal":{"ja":"Proc. of the 25th International Conference on Pattern Recognition (ICPR2020)"},"pages":"8766-8773","year":2021,"month":1,"location":"Online","review":"reviewed","fields":["pattern-recognition"],"doi":"10.1109/ICPR48806.2021.9413144"},{"key":"Minatani2021","id":"1849","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"南谷 和範","en":""},{"ja":"渡辺 哲也","en":""},{"ja":"岩村 雅一","en":""}],"title":{"ja":"視覚障害者を対象とした3Dモデル触察シンポジウムのオンライン開催の試み"},"journal":{"ja":"日本図学会 2021年度大会 講演論文集"},"year":2021,"month":11,"fields":["visually-impaired-assistance"]},{"key":"Nakatani2021_MIRU","id":"1847","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"中谷 友哉","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"藤本 仰一","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"イチリンソウとその近縁種における隣接花弁の重なり順推定"},"journal":{"ja":"第24回 画像の認識・理解シンポジウム（MIRU2021）予稿集"},"presen_id":"I21-11","num_pages":"4 pages","year":2021,"month":7,"review":"not_reviewed","fields":[]},{"key":"Nishida2021","id":"1837","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"西田龍平","en":""},{"ja":"内海ゆづ子","en":""},{"ja":"三輪由佳","en":""},{"ja":"岩村雅一","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"ブドウ摘房支援のための画像における房計数領域の推定"},"series":"情報処理学会研究報告コンピュータビジョンとイメージメディア（CVIM）","presen_id":"2020-CVIM-226,43","pages":"1-7","year":2021,"month":5,"location":"オンライン開催","review":"not_reviewed","fields":[]},{"key":"Omura2021_CVIM","id":"1838","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"大村 海雲","en":"Kaiun Omura"},{"ja":"内海 ゆづ子","en":"Yuzuko Utsumi"},{"ja":"三輪 由佳","en":"Yuka Miwa"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"ブドウ果粒検出の評価用画像データセットの構築"},"journal":{"ja":"研究報告コンピュータビジョンとイメージメディア（CVIM）","en":"Computer Vision and Image Media"},"volume":"2021-CVIM-226","presen_id":"35","pages":"1-5","num_pages":"5","year":2021,"month":5,"location":"オンライン開催","review":"not_reviewed","fields":["object-recognition","plant-image-processing"],"url":"https://ipsj.ixsq.nii.ac.jp/ej/?action=pages_view_main&active_action=repository_view_main_item_detail&item_id=211186&item_no=1&page_id=13&block_id=8","abstract":{"ja":"日本のブドウ生産は年々減少している．この原因として，高齢化による栽培者の減少とブドウ栽培作業の数の多さ，困難さが挙げられる．そこで，我々は，ブドウの栽培支援を目的として，摘粒支援システムを開発している．このシステムの評価のためには，画像上からの実の検出位置精度の評価が必要であるが，ブドウの実の画像とその位置を評価するためのデータセットは我々の知る限り存在していない．そのため，本研究では，摘粒支援システムの評価用データセットを作成する．このデータセットは，3210枚のブドウ画像とブドウの実の中心位置情報から成り，ブドウの画像から実の中心位置を検出する手法を評価できる．データセットの難易度を評価するため，従来研究でブドウの実の検出に成功していた深層学習以前に提案されていたRoscherらの手法を適用し，ブドウの実の中心位置検出の精度評価を行った．実験の結果，一部のブドウの実のみ検出することができたが，多くのブドウの実の検出には失敗した．このことから，提案データベースのタスクは，深層学習以前の手法に対して難しいことが明らかとなった．"}},{"key":"Nawa2021_GC","id":"1823","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"名和 俊","en":""},{"ja":"Islam Md. Rabiul","en":""},{"ja":"Andrew Vargo","en":""},{"ja":"岩田 基","en":""},{"ja":"松原 正樹","en":""},{"ja":"森嶋 厚行","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"視線情報に基づいたクラウドワークの品質評価"},"journal":{"ja":"2021年 電子情報通信学会 総合大会　論文集"},"presen_id":"D4-4","num_pages":"1","year":2021,"month":3,"fields":["crowdsourcing"]},{"key":"Kawai2021_WS","id":"1810","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"河合 隆哲","en":""},{"ja":"高嶋 慶伍","en":""},{"ja":"岩村 雅一","en":""},{"ja":"南谷 和範","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"視覚障害者の受動的な情報取得に向けての情報要約に関する調査"},"journal":{"ja":"第181回ヒューマンインタフェース学会研究会"},"volume":"23","number":"1","pages":"5-14","year":2021,"month":3,"review":"not_reviewed","presen_type":"oral","fields":["visually-impaired-assistance","pattern-recognition"]},{"key":"Takashima2021_WS","id":"1811","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"高嶋 慶伍","en":""},{"ja":"河合 隆哲","en":""},{"ja":"岩村 雅一","en":""},{"ja":"南谷 和範","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"視覚障害者の周囲の視覚情報取得におけるQAシステムの有用性に関する調査"},"journal":{"ja":"第181回ヒューマンインタフェース学会研究会"},"volume":"23","number":"1","pages":"15-24","year":2021,"month":3,"review":"not_reviewed","presen_type":"oral","fields":["visually-impaired-assistance","pattern-recognition"]},{"key":"Takaike2021","id":"1808","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"高池 太郎","en":""},{"ja":"岩田 基","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"視線情報と座圧情報を用いた多読時の漫画英訳本に対する興味度推定"},"journal":{"ja":"第5回コミック工学研究会予稿集"},"pages":"72-76","year":2021,"month":3,"review":"not_reviewed","fields":[]},{"key":"Tsuruta2021","id":"1809","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"鶴田 あおい","en":""},{"ja":"岩田 基","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"陰影付き線画を用いた光源位置推定"},"journal":{"ja":"第5回コミック工学研究会予稿集"},"pages":"111-114","year":2021,"month":3,"review":"not_reviewed","fields":[]},{"key":"Maruyama2021","id":"1813","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"丸山 瑞己","en":""},{"ja":"Shuvozit Ghose","en":""},{"ja":"井上 勝文","en":""},{"ja":"Partha Pratim Roy","en":""},{"ja":"岩村 雅一","en":""},{"ja":"吉岡 理文","en":""}],"title":{"ja":"局所領域に着目したMulti-stream Neural Networks による手話単語認識"},"journal":{"ja":"電子情報通信学会技術研究報告 パターン認識・メディア理解研究会(PRMU)"},"year":2021,"month":3,"review":"not_reviewed","fields":["pattern-recognition"],"url":"https://www.ieice.org/ken/paper/20210304FC2a/","awards":{"ja":["2021年3月の月間ベストプレゼンテーション賞受賞"],"en":["March 2021 PRMU Best Presentation Award Winner"]}},{"key":"Nakatani2021_PRMU","id":"1807","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"中谷 友哉","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"藤本 仰一","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"多種少量データを対象とした隣接花弁の重なり順推定"},"journal":{"ja":"電子情報通信学会技術研究報告 パターン認識・メディア理解研究会(PRMU)"},"volume":"120","number":"409","pages":"115-120","year":2021,"month":3,"review":"not_reviewed","presen_type":"oral_and_poster","fields":["plant-image-processing","object-recognition"],"url":"https://www.ieice.org/ken/paper/20210305OC2A/"},{"key":"Islam2021_HCI","id":"1815","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"Md. Rabiul Islam","en":"Md. Rabiul Islam"},{"ja":"Shuji Sakamoto","en":"Shuji Sakamoto"},{"ja":"Yoshihiro Yamada","en":"Yoshihiro Yamada"},{"ja":"Andrew Vargo","en":"Andrew Vargo"},{"ja":"Motoi Iwata","en":"Motoi Iwata"},{"ja":"Masakazu Iwamura","en":"Masakazu Iwamura"},{"ja":"Koichi Kise","en":"Koichi Kise"}],"title":{"ja":"Reading Activity Classification Using Self-supervised Deep Learning","en":"Reading Activity Classification Using Self-supervised Deep Learning"},"journal":{"ja":"研究報告ヒューマンコンピュータインタラクション（HCI）"},"volume":"2021-HCI-191","number":"21","pages":"1-8","year":2021,"month":1,"location":"オンライン","fields":[]},{"key":"Oota2021","id":"1885","type":"survey","language":"JA","show":"show","authors":[{"ja":"太田智美","en":""}],"title":{"ja":"座談会：ゴール設定の違いが難しさ　「情報植物学」という新たな分野への挑戦"},"book_title":"情報処理","volume":"62","number":"12","year":2021,"month":11,"review":"not_reviewed","fields":["plant-image-processing"],"url":"https://note.com/ipsj/n/na92a86b9f405?magazine_key=m44cc2603aaa9","note":{"ja":"先行記事．内海が座談会に参加"}},{"key":"Utsumi2021","id":"1884","type":"survey","language":"JA","show":"show","authors":[{"ja":"内海 ゆづ子","en":""}],"title":{"ja":"画像認識分野から見た植物の研究"},"journal":{"ja":"情報処理"},"volume":"62","number":"12","num_pages":"3","year":2021,"month":11,"review":"not_reviewed","fields":["plant-image-processing"]},{"key":"Kise2021b","id":"1876","type":"other","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""}],"title":{"ja":"Reading of \"Reading\" for \"Actuating\": Augmenting Human Learning by \"Experiential Supplements"},"journal":{"ja":"The 3rd International Conference on Activity and Behavior Computing"},"year":2021,"month":10,"review":"not_reviewed","fields":["activity-recognition","learning-analytics"],"note":{"ja":"Keynote talk"}},{"key":"Kise2021a","id":"1875","type":"other","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""}],"title":{"ja":"Beyond Camera-Based Document Analysis and Recognition: Towards Camera-Based Human Document Interactions"},"journal":{"ja":"International Workshop on Camera-Based Document Analysis and Recognition"},"year":2021,"month":9,"review":"not_reviewed","presen_type":"oral","fields":["activity-recognition","document-image-analysis","learning-analytics"],"note":{"en":"Keynote talk"}},{"key":"Shippitsu2021","id":"1927","type":"other","language":"JA","show":"show","authors":[{"ja":"執筆：太田智美","en":""},{"ja":"参加：中島敬二","en":""},{"ja":"稲見昌彦","en":""},{"ja":"上田貴志","en":""},{"ja":"植田美那子","en":""},{"ja":"近藤洋平","en":""},{"ja":"内海ゆづ子","en":""},{"ja":"太田智美","en":""}],"title":{"ja":"座談会：ゴール設定の違いが難しさ「情報植物学」という新たな分野への挑戦"},"journal":{"ja":"情報処理"},"volume":"62","number":"12","year":2021,"month":9,"review":"not_reviewed","fields":["plant-image-processing"],"url":"https://note.com/ipsj/n/na92a86b9f405","note":{"ja":"note 先行公開記事"}},{"key":"Utsumi2021_Conf","id":"1882","type":"other","language":"JA","show":"show","authors":[{"ja":"内海 ゆづ子","en":""}],"title":{"ja":"コンピュータビジョン・パターン認識技術に基づく画像を用いた植物の空間的周期性の解析"},"journal":{"ja":"日本植物学会第85回大会"},"year":2021,"month":9,"review":"not_reviewed","presen_type":"oral","fields":["plant-image-processing"],"note":{"ja":"招待講演"}},{"key":"Ishimaru2021","id":"1840","type":"other","language":"EN","show":"show","authors":[{"ja":"Shoya Ishimaru","en":""},{"ja":"Takanori Maruichi","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Confidence-Aware Learning Assistant"},"journal":{"ja":"arXiv:2102.07312 [cs.HC]"},"year":2021,"month":2,"review":"not_reviewed","fields":["learning-analytics","sensing"]},{"key":"Kajopoulos2020","id":"1754","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Jasmin Kajopoulos","en":""},{"ja":"Gordon Cheng","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Hermann J. Müller","en":""},{"ja":"Agnieszka Wykowska","en":""}],"title":{"ja":"Focusing on the face or getting distracted by social signals? The effect of distracting gestures on attentional focus in natural interaction"},"journal":{"ja":"Psychological Research"},"year":2020,"month":8,"review":"reviewed","fields":["sensing"],"doi":"10.1007/s00426-020-01383-4","abstract":{"ja":"Attentional orienting towards others’ gaze direction or pointing has been well investigated in laboratory conditions. However, less is known about the operation of attentional mechanisms in online naturalistic social interaction scenarios. It is equally plausible that following social directional cues (gaze, pointing) occurs reflexively, and/or that it is influenced by top-down cognitive factors. In a mobile eye-tracking experiment, we show that under natural interaction conditions, overt attentional orienting is not necessarily reflexively triggered by pointing gestures or a combination of gaze shifts and pointing gestures. We found that participants conversing with an experimenter, who, during the interaction, would play out pointing gestures as well as directional gaze movements, continued to mostly focus their gaze on the face of the experimenter, demonstrating the significance of attending to the face of the interaction partner?in line with effective top-down control over reflexive orienting of attention in the direction of  social cues."}},{"key":"Miyazaki2020","id":"1685","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Tomo Miyazaki","en":""},{"ja":"Tatsunori Tsuchiya","en":""},{"ja":"Yoshihiro Sugaya","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Automatic Generation of Typographic Font from Small Font Subset"},"journal":{"ja":"IEEE Computer Graphics and Applications"},"volume":"40","number":"1","pages":"99-111","year":2020,"month":1,"review":"reviewed","fields":["document-image-analysis"],"doi":"10.1109/MCG.2019.2931431"},{"key":"Utsumi2020_JST","id":"1806","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Yuzuko Utsumi","en":""}],"title":{"ja":"DNN-based plant phenome estimation with small data"},"journal":{"ja":"2020 JST The Second International Workshop on Field Phenotyping and Modeling for Cultivation"},"year":2020,"month":12,"location":"Online","fields":["plant-image-processing","pattern-recognition"]},{"key":"Kawakami2020","id":"1804","type":"intl_conf","language":"JA","show":"show","authors":[{"ja":"Rei Kawakami","en":""},{"ja":"Ikuo Kamei","en":""},{"ja":"Changyo Han","en":""},{"ja":"Takefumi Hiraki","en":""},{"ja":"Yuzuko Utsumi","en":""},{"ja":"Yasuaki Kakehi","en":""},{"ja":"Takeshi Naemura","en":""}],"title":{"ja":"OUR Shurijo: Shurijo Castle Digital Reconstruction Project"},"journal":{"ja":"International Conference on 3D Vision"},"presen_id":"7","year":2020,"month":11,"location":"Online","review":"not_reviewed","presen_type":"poster","fields":["3d-reconstruction"]},{"key":"Iwamura2020_ICCHP","id":"1753","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Yoshihiko Inoue","en":""},{"ja":"Kazunori Minatani","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Suitable Camera and Rotation Navigation for People with Visual Impairment on Looking for Something Using Object Detection Technique"},"journal":{"ja":"Proc. 17th International Conference on Computers Helping People with Special Needs (ICCHP 2020)"},"pages":"495-509","year":2020,"month":9,"location":"Online","review":"reviewed","presen_type":"oral","fields":["visually-impaired-assistance","object-recognition"],"doi":"10.1007/978-3-030-58796-3_57","note":{"ja":"Since the initial version of the paper has a big bug, please download the latest version."},"abstract":{"ja":"For people with visual impairment, smartphone apps that use computer vision techniques to provide visual information have played important roles in supporting their daily lives.\n  However, they can be used under a specific condition only.\n  That is, only when the user knows where the object of interest is.\n  In this paper, we first point out the fact mentioned above by categorizing the tasks that obtain visual information using computer vision techniques.\n  Then, in looking for something as a representative task in a category,\n  we argue suitable camera systems and rotation navigation methods.\n  In the latter, we propose novel voice navigation methods.\n  As a result of a user study comprised of seven people with visual impairment, we found that (1) a camera with a wide field of view such as an omnidirectional camera was preferred, and (2) users have different preferences in navigation methods."}},{"key":"Yamaguchi2020_ISWC","id":"1755","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"","en":"Kohei Yamaguchi"},{"ja":"","en":"Motoi Iwata"},{"ja":"","en":"Andrew Vargo"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Mobile Vocabulometer: A Context-based Learning Mobile Application to Enhance English Vocabulary Acquisition"},"journal":{"en":"Proceedings of the 2020 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of the 2020 ACM International Symposium on Wearable Computers (UbiComp/ISWC '20 Adjunct)"},"num_pages":"4","year":2020,"month":9,"location":"UbiComp/ISWC ’20","review":"reviewed","presen_type":"oral_and_poster","fields":["reading-life-log"],"doi":"10.1145/3410530.3414406","url":"https://doi.org/10.1145/3410530.3414406","abstract":{"en":"Vocabulary acquisition is the basis of learning a language, and using flashcards applications is a popular method for learners to memorize the meaning of unknown words. Unfortunately, this method alone is not effective for learners to remember the meaning of words when they appear in sentences. To solve this, we developed the Mobile Vocabulometer which allows users to acquire new vocabulary with context-based learning. Based on the correlation between comprehension and interests, we use the learning materials that adapt to users' interests and language skills. This system harnesses the power of the original Vocabulometer, and modifies it to be effective for mobile learning. An experiment on Japanese university students showed that, overall, learners achieved better results compared to using a simple flashcard application. This result indicates that this system provides a significant advantage over context-free learning systems."}},{"key":"Muta2020","id":"1760","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Kaoru Muta","en":"Kaoru Muta"},{"ja":"Shiho Takada","en":"Shiho Takada"},{"ja":"Yuzuko Utsumi","en":"Yuzuko Utsumi"},{"ja":"Atsushi Matsumura","en":"Atsushi Matsumura"},{"ja":"Masakazu Iwamura","en":"Masakazu Iwamura"},{"ja":"Koichi Kise","en":"Koichi Kise"}],"title":{"ja":"Automatic Calculation of Infection Rate of Arbuscular Mycorrhizal Fungi Using Deep CNN","en":"Automatic Calculation of Infection Rate of Arbuscular Mycorrhizal Fungi Using Deep CNN"},"journal":{"ja":"ECCV 2020 workshop on Computer Vision Problems in Plant Phenotyping (CVPPP 2020)","en":"ECCV 2020 workshop on Computer Vision Problems in Plant Phenotyping (CVPPP 2020)"},"presen_id":"3","num_pages":"1","year":2020,"month":8,"location":"Online","review":"reviewed","presen_type":"poster","fields":["plant-image-processing","pattern-recognition","deep-learning","object-recognition"],"url":"https://www.plant-phenotyping.org/lw_resource/datapool/systemfiles/elements/files/9f71f5db-eb90-11ea-908c-dead53a91d31/current/document/03MutaCVPPP2020.pdf"},{"key":"Miyamoto2020_COCOON","id":"1752","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masayuki Miyamoto","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""},{"ja":"François Le Gall","en":""}],"title":{"ja":"Quantum Speedup for the Minimum Steiner Tree Problem"},"journal":{"ja":"Proc. 26th International Computing and Combinatorics Conference (COCOON 2020)"},"pages":"234-245","year":2020,"month":8,"location":"Online","review":"reviewed","fields":["quantum-computing"],"doi":"10.1007/978-3-030-58150-3_19"},{"key":"Iwamura2020_CHI","id":"1712","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Naoki Hirabayashi","en":""},{"ja":"Zheng Cheng","en":""},{"ja":"Kazunori Minatani","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"VisPhoto: Photography for People with Visual Impairment as Post-Production of Omni-Directional Camera Image"},"journal":{"ja":"ACM CHI Conference on Human Factors in Computing Systems (CHI2020) Extended Abstracts Proceedings"},"pages":"1-9","year":2020,"month":4,"location":"(The conference was not held due to the COVID-19 pandemic)","review":"reviewed","presen_type":"poster","fields":["visually-impaired-assistance"],"doi":"10.1145/3334480.3382983","note":{"ja":"Preview video is available at <a href=\"https://youtu.be/hOZB-Anhs18\">https://youtu.be/hOZB-Anhs18</a>"},"abstract":{"ja":"It is known that many people with visual impairments desire to take photos. In taking photos, however, they often face difficulty in pointing the camera at the subject. In this paper, we address this problem by proposing a novel photo-taking system named VisPhoto. Unlike existing ones, the proposed system generates a photo as post-production of an omni-directional camera image. That is, when the shutter button is pressed, the proposed system captures an omni-directional camera image. In the post-production, from the objects detected in the image, the user selects the ones that should be included and excluded. Finally, as a ``photo,'' the system outputs a cropped region satisfying the user's preference and looking beautiful from an aesthetic point of view."}},{"key":"Tonomoto2020","id":"1744","type":"intl_conf","language":"JA","show":"show","authors":[{"ja":"Sadafumi Tonomoto","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Preliminary Experiments Toward Personalized Nudging Strategies Extensive Reading of English"},"journal":{"ja":"Proc. of CHI2020 Workshop on Detection and Design for Cognitive Biases in People and Computing Systems"},"num_pages":"8","year":2020,"month":4,"review":"reviewed","fields":["learning-analytics"],"abstract":{"ja":"Extensive reading is a way of improving language ability by just reading a lot. Although there are some recommended rules to follow, it is sometimes difficult to sustain the user’s engagement in reading. It would be helpful to have the technology for sustainment. In this paper we report our first trial to establish nudging strategies for this purpose. From the experiments with 29 participants for 25 days in total, we have found that setting the goal of the amount of reading, as well as sharing the goal with a peer group of users with a similar ability is effective nudging strategies. Although the goal setting is not effective to the majority of users, we have been successful to prepare a prescription by which users with positive effects are selected based on their personal traits and the amount of reading. From the results, we are able to approach to personalized nudging strategies."}},{"key":"Mahad2020","id":"1736","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Fairuz Safwan Mahad","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Leveraging Pyramidal Feature Hierarchy for 3D Reconstruction"},"journal":{"ja":"Proc. International Workshop on Frontiers of Computer Vision (IW-FCV)"},"pages":"347-362","year":2020,"month":2,"location":"Ibusuki, Japan","review":"reviewed","presen_type":"oral","fields":["3d-reconstruction"],"doi":"10.1007/978-981-15-4818-5_26","abstract":{"ja":"3D reconstruction methods with neural networks has been on a rising trend and extensive research has been done. However, they do not pay particular attention to reconstructing the detailed parts of the objects. Although the reconstructed 3D models often look complete generally, close-up observations reveal that most of them lose out on the detailed parts, which reduces the quality of the 3D model. We observe that especially fine parts of the object either fail to be reconstructed or are less dense and incomplete. This happens because the network is not designed to capture the fine details of the object. Therefore, in this paper, we propose a network which is designed to capture both the coarse and fine details of the object in an effort to improve the reconstruction of the fine parts of the object. In order to do so, we design a multi-branch deep generative network which learns the local features, generic features and also intermediate features. Utilizing these features allows the network to learn features at different levels which are able to reconstruct the fine parts as well as the overall shape of the reconstructed 3D model. We show that our method outperformed the state-of-the-art."}},{"key":"Watanabe2020","id":"1792","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"渡辺 哲也","en":""},{"ja":"南谷 和範","en":""},{"ja":"岩村 雅一","en":""}],"title":{"ja":"視覚障害者への3D模型提供サービス — サービス利用とアンケートの分析 —"},"journal":{"ja":"電子情報通信学会技術研究報告、WIT HI-SIGACI"},"year":2020,"month":12,"review":"not_reviewed","presen_type":"oral","fields":["visually-impaired-assistance"],"url":"https://www.ieice.org/ken/paper/20201210yCBk/"},{"key":"Miyamoto2020","id":"1793","type":"domestic_conf","language":"EN","show":"show","authors":[{"ja":"Masayuki Miyamoto","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""},{"ja":"François Le Gall","en":""}],"title":{"ja":"Quantum Speedup for the Minimum Steiner Tree Problem"},"journal":{"ja":"情報処理学会 第1回量子ソフトウェア研究発表会"},"year":2020,"month":10,"review":"not_reviewed","fields":["quantum-computing"]},{"key":"Akai2020_MIRU","id":"1756","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"赤井 亮太","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"三輪 由佳","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"全方位画像を用いたブドウの房数推定―歪みに応じた密度マップの生成とアラインメント―"},"journal":{"ja":"第23回 画像の認識・理解シンポジウム(MIRU2020)予稿集"},"presen_id":"OS2-3B-3","num_pages":"4","year":2020,"month":8,"review":"reviewed","presen_type":"oral_and_poster","fields":["plant-image-processing","pattern-recognition"],"note":{"ja":"オンライン開催"}},{"key":"Akai2020_CVIM","id":"1747","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"赤井 亮太","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"三輪 由佳","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"全方位カメラ画像における歪みを考慮したブドウの房数推定"},"journal":{"ja":"情報処理学会研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2020-CVIM-222","number":"36","pages":"1-7","year":2020,"month":5,"location":"オンライン開催","review":"not_reviewed","fields":["plant-image-processing","deep-learning"],"url":"http://id.nii.ac.jp/1001/00204373/"},{"key":"Kise2020","id":"1748","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"黄瀬 陸哉","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"Self-supervised learning を用いた画像からの単子葉植物の分げつ数推定"},"journal":{"ja":"情報処理学会研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2020-CVIM-222","number":"35","pages":"1-7","year":2020,"month":5,"location":"オンライン開催","review":"not_reviewed","fields":["plant-image-processing","deep-learning"],"url":"http://id.nii.ac.jp/1001/00204372/"},{"key":"Utsumi2020_Conf","id":"1758","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"内海 ゆづ子","en":""},{"ja":"赤井 亮太","en":""},{"ja":"三輪 由佳","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"摘房支援のためのCNNを用いた回帰に基づくブドウの房数推定"},"journal":{"ja":"農業情報学会2020年年度年次大会予稿集"},"presen_id":"C-06","num_pages":"86-87","year":2020,"month":5,"fields":["plant-image-processing","pattern-recognition"]},{"key":"Iwamura2020b","id":"1741","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村雅一","en":""},{"ja":"平林直樹","en":""},{"ja":"程征","en":""},{"ja":"南谷和範","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"VisPhoto: 全方位カメラを用いた視覚障碍者のための写真撮影支援システム"},"journal":{"ja":"インタラクション2020論文集"},"presen_id":"3B-18","pages":"874-879","num_pages":"6","year":2020,"month":3,"location":"インタラクション2020","review":"not_reviewed","fields":["visually-impaired-assistance","deep-learning","object-recognition"],"abstract":{"ja":"多くの視覚障碍者が晴眼者（目が見える人）と同じように写真を撮りたいこと\nが知られている．写真を通して「印象に残る場面を画像で残したい」，「体験を他人\nと共有して感動を分かち合いたい」という需要は晴眼者と同じである．しかし，視\n覚障碍者，特に全盲の人達は被写体が見えないため，写したいもの全体をフレー\nムに入れることが難しい．この問題点を解決するために，リアルタイムで処理が\n可能なアプリケーションが開発されているが，被写体が静止しているか動きが遅\nい場合にしか使えない．\n本稿では上記の問題を解決するために全方位カメラを用いる革新的な写真撮影\n手法 VisPhoto を紹介する．提案システムでは，写真撮影時に全方位カメラを用い\nて撮影者の周囲の景色すべてを一度に撮影する．その後，サーバ上で撮影した全\n方位画像から物体が検出され，その情報を手がかりにユーザは望んだ領域を切り\n出す．その際，写真としてより美しい領域を切り取る．全方位カメラを用いて瞬\n時に周囲を撮影してしまい，後処理でユーザの望む領域を切り取ることから，既\n存手法のようにカメラを被写体に向けるために時間をかける必要がない．"}},{"key":"Taniguchi2020","id":"1740","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"谷口 善信","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"タッチレスなタッチパネル操作デバイスの提案"},"journal":{"ja":"インタラクション2020論文集"},"presen_id":"3P-79","pages":"1070-1073","num_pages":"4","year":2020,"month":3,"location":"インタラクション2020","review":"not_reviewed","fields":["visually-impaired-assistance"],"note":{"ja":"3P-79_revised.pdf(3P-79の修正版.pdf)\n修正内容\n2ページ　図2，図3の出典を表記\n4ページ　3.3節　「ウェブインターフェー上ス」から「ウェブインターフェース上」"},"abstract":{"ja":"省スペース化や操作のしやすさから，タッチパネルを搭載した機器が増加している．視覚障碍者にとってタッチパネルは画面からの情報を受け取ることができないため，晴眼者に比べて操作時にかかる負担が大きく，何らかの支援が必要であると考えられる．しかし，視覚障碍者に画面情報と画面推移情報を伝え，ボタン位置への誘導を行おうとすると，画面から情報をきちんと読み取り伝えなければならないうえに，時間もかかり，誤タッチをしてしまう可能性もある．そこで本研究では，画面情報と画面推移情報をあらかじめ保持しボタン押下機構を持ったタッチパネル操作デバイスを提案する．"}},{"key":"Oshiba2020_WSb","id":"1816","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"大芝 潤気","en":"Junki Oshiba"},{"ja":"岩田 基","en":"Motoi Iwata"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"C2GANを用いたランドマークに基づくアニメキャラクタの顔画像自動生成手法"},"journal":{"ja":"第3回コミック工学研究会"},"year":2020,"month":3,"review":"not_reviewed","presen_type":"oral","fields":["comic-computing","deep-learning"],"url":"https://drive.google.com/file/d/18djOH5bNTPRTwC3j1X7NM50yN0Dc8er6/view","abstract":{"ja":"近年、アニメーションが付与された「動く漫画」が登場している。しかしながらアニメーションの作成には手間と時間がかかるため、「動く漫画」化されているのは一部の作品に限られている。本研究では、表情変化のアニメーションを自動生成することを目的とし、C2GANを用いてランドマークに基づく顔画像を生成する手法を提案する。アニメ動画から作成したデータセットを用いてC2GANを学習することで提案を実現し、提案手法の有効性を示す。"}},{"key":"Oshiba2020_WSa","id":"1738","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"大芝 潤気","en":""},{"ja":"岩田 基","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"C2GANを用いたランドマークに基づくアニメキャラクタの顔画像自動生成手法"},"journal":{"ja":"電子情報通信学会 ヒューマンコミュニケーショングループ 第二種時限研究会 第3回コミック工学研究会"},"num_pages":"6","year":2020,"month":3,"location":"大阪府 I-siteなんば","review":"not_reviewed","fields":["comic-computing","deep-learning"]},{"key":"Iwamura2020a","id":"1713","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":""},{"ja":"平林 直樹","en":""},{"ja":"程 征","en":""},{"ja":"南谷 和範","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"全方位カメラを用いた視覚障碍者のための写真撮影支援システム VisPhoto"},"journal":{"ja":"情報アクセシビリティをめぐる諸問題に関する研究集会"},"year":2020,"month":2,"review":"not_reviewed","presen_type":"oral","fields":["visually-impaired-assistance"]},{"key":"Minatani2020","id":"1716","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"南谷 和範","en":""},{"ja":"渡辺 哲也","en":""},{"ja":"岩村 雅一","en":""}],"title":{"ja":"視覚障害者のための3D造形物配信・出力エコシステム構築に向けて"},"journal":{"ja":"情報アクセシビリティをめぐる諸問題に関する研究集会"},"year":2020,"month":2,"fields":["visually-impaired-assistance"]},{"key":"Donomoto2020_HCI","id":"1715","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"殿元 禎史","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"英文多読のための個人に適合したナッジ戦略の検討"},"journal":{"ja":"情報処理学会研究報告ヒューマンコンピュータインタラクション (HCI)"},"volume":"2020-HCI-186","number":"9","series":"ヒューマンコンピュータインタラクション(HCI)","pages":"1-6","num_pages":"6","year":2020,"month":1,"location":"石垣島 大濱信泉記念館","review":"not_reviewed","fields":["reading-life-log","learning-analytics"],"abstract":{"ja":"集中度や確信度など，学習のパフォーマンスを示す指標を推定する研究は多々存在する．しかし，どのように結果を利用すればパフォーマンスが向上するのかを検証した研究は少ない．また，そのような研究は統計として全体に効果があるものとしているが，個人によって物事の捉え方は異なるためパフォーマンスの改善には差がある．そこで，人の心理的傾向を定量化することで，最も効果を発揮する個人に適合した学習戦略を構築することを検討する．本研究では，戦略として学習意欲を継続させるため，行動経済学において人に強制することなく自発的な行動を促すナッジを用いたナッジ戦略を検討する．また，英語学習の代表的なものとして多読を対象に実験を行った．多読とは，単語や文法の学習とは異なり，英語を英語のまま理解できるよう，多く英文を読むことである．実験では，13 名を対象に 20 日間英文読書を行ってもらい，実験期間中にフィードバックを与え，読書量が伸びるかを検証した．結果として，目標設定が有効な人は開放性が高く神経症的傾向が低い可能性があること，既に読書習慣を持つ人は開放性が極端に低く誠実性が高い可能性があることがわかった．心理的傾向から読書量を増加させるのに有効なフィードバックが分類できる可能性を示唆した．また，選択する文書の難易度を心理的傾向から分類できる可能性も示唆した．"}},{"key":"Maruichi2020_HCI","id":"1743","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"丸市賢功","en":"Takanori Maruichi"},{"ja":"黄瀬浩一","en":"Koichi Kise"}],"title":{"ja":"対数正規分布を用いた英単語筆記時の確信度推定手法の提案"},"journal":{"ja":"情報処理学会研究報告ヒューマンコンピュータインタラクション (HCI)"},"volume":"2020-HCI-186","number":"9","series":"ヒューマンコンピュータインタラクション(HCI)","num_pages":"7","year":2020,"month":1,"location":"HCI研究会@石垣島 大濱信泉記念館","review":"not_reviewed","fields":["pattern-recognition","learning-analytics"],"abstract":{"ja":"英単語の学習において，復習は重要なプロセスである．一般的に復習は不正解の英単語に対して行われる．しかし，この方法では確信なくたまたま正解してしまった英単語を復習することができない．\nそこで本研究では，このような英単語を復習に含めるため，解答に関する確信度を推定する手法を提案する．これにより，確信をもって不正解した英単語に特別の注意を与えることも可能となる．\n本研究では，確信度推定のために使用する学習者の振舞いとして，英単語問題解答時の手書きログから抽出した特徴量に着目する．これらの特徴量の分布は対数正規分布に従うことが確認されたため，対数正規分布を用いて確信度推定を行う手法を提案する．また，提案手法を推定精度を検証するために21名を対象に実験を行い，実験の精度について正規分布で仮定したものとの比較を行った．各手法のユーザ依存での平均推定精度は，正規分布を仮定した場合では76%，対数正規分布を仮定した場合では78%であった．"}},{"key":"Yuasa2020_HCI","id":"1714","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"湯浅 成章","en":"Shigeaki Yuasa"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"有酸素運動が英単語暗記に及ぼす影響の確認"},"journal":{"ja":"情報処理学会研究報告ヒューマンコンピュータインタラクション (HCI)"},"volume":"2020-HCI-186","number":"9","series":"ヒューマンコンピュータインタラクション(HCI)","num_pages":"5","year":2020,"month":1,"location":"HCI研究会@石垣島 大濱信泉記念館","review":"not_reviewed","presen_type":"oral","fields":["learning-analytics"],"abstract":{"ja":"我々は，学習増強を目指して，これまでに確信度に基づいた復習や，目標設定を利用した読書量の増加などの学習法を提案している．本研究では，これらに加わる新しい学習増強の方法として，有酸素運動を利用した英単語暗記法を提案する．従来の研究から，有酸素運動が脳の認知機能を改善することが明らかになっている．本研究は，この知見に基づくものである．有酸素運動の方法としては，場所を取らず，危険性も少ないステッパを用いる．英単語の暗記は，スマートフォンのフラッシュカードアプリを用いて行う．本手法の有効性を検証するために，実験参加者10名を対象として実験を行った．暗記の1日後，3日後，1週間後にテストを行った結果，1日後には有酸素運動による差が認められなかったものの，3日後では7パーセント，1週間後では6パーセント，正答率が向上した．また，この差は統計的に有意であることを確認した．"}},{"key":"Yamada2020","id":"1719","type":"doctoral_thesis","language":"JA","show":"show_except_abstract","authors":[{"ja":"山田 良博","en":""}],"title":{"ja":"ニューラルネットワークの確率的正則化に関する研究"},"year":2020,"month":2,"fields":[]},{"key":"Imagawa2020_IL","id":"1750","type":"survey","language":"JA","show":"show_except_datafile","authors":[{"ja":"今川 寛之","en":"Hiroyuki IMAGAWA"},{"ja":"岩田 基","en":"Motoi IWATA"},{"ja":"黄瀬 浩一","en":"Koichi KISE"}],"title":{"ja":"深層学習を利用した印刷物用電子透かし法の実現に向けて","en":"Toward the realization of a digital watermarking method against print-cam attack using deep learning"},"journal":{"ja":"画像ラボ","en":"Image Laboratory"},"volume":"31","number":"6","pages":"67-72","num_pages":"6","year":2020,"month":6,"publisher":"日本工業出版","review":"not_reviewed","fields":["digital-watermarking"]},{"key":"Iwamura2020_J","id":"1751","type":"survey","language":"JA","show":"show","authors":[{"ja":"岩村雅一","en":""},{"ja":"平林直樹","en":""},{"ja":"程征","en":""},{"ja":"南谷和範","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"全方位カメラを用いた視覚障碍者の写真撮影支援"},"journal":{"ja":"知能と情報（日本知能情報ファジィ学会誌）"},"volume":"32","number":"3","pages":"80-86","year":2020,"month":6,"fields":["visually-impaired-assistance","deep-learning","object-recognition"],"abstract":{"ja":"本稿では，全盲の視覚障碍者が独力で写真を撮れる全く新しいシステムを紹介する．\n従来の写真撮影支援システムは，写真撮影時に被写体がフレーム内に入るように，もしくはより良い写真が撮れるように，写真撮影に用いるスマートフォンの向きを音声や振動で利用者に指示するものであった．\nそれに対して，提案手法 VisPhoto は，全方位カメラがカメラの周囲360度の景色を撮影できることを利用して，利用者の周囲を全方位画像として撮影して，その後処理として利用者の希望を反映した「写真」を切り取る．\n後処理では，全方位画像に物体検出手法を適用して得られた画像中の物体の種類と方向を利用者に提示し，「写真」に含めたい物体を入力してもらう．\n「写真」を切り出す際には，利用者の希望を反映させるとともに，写真の見た目も良くなるように一工夫している．\n提案手法は，システムの有用性，操作性を8名の全盲の視覚障碍者に，\n撮影した「写真」の質を10名の晴眼者（目が見える人）に検証してもらい，いずれも肯定的な評価を得た．全方位カメラを用いた視覚障碍者の写真撮影支援"}},{"key":"Iwamura2020_TELECOM","id":"1746","type":"survey","language":"JA","show":"show","authors":[{"ja":"岩村雅一","en":"Masakazu Iwamura"}],"title":{"ja":"k近傍探索のための量子アルゴリズム","en":"A Quantum Algorithm for k-Nearest Neighbor Search"},"journal":{"ja":"TELECOM FRONTIER","en":"TELECOM FRONTIER"},"volume":"107","year":2020,"month":5,"review":"not_reviewed","fields":["quantum-computing","nearest-neighbor-search"],"url":"http://www.scat.or.jp/frontier/frontier107/mokuji_107.html","abstract":{"ja":"k近傍探索とは、検索質問データが与えられたとき、それに最も近いk個のデータをデータベースから探し出す問題である。この問題は、時間を掛けさえすれば必ず解ける。しかし、データベースの大規模化や高次元化に伴い、計算時間が大きくなってしまう。この問題を高速に解くために、探索過程に近似を導入し、探索誤りを許容するのが近似k近傍探索である。本研究では、従来用いられている計算機（古典的コンピュータ）ではなく、量子コンピュータで動作するk近傍探索手法の量子アルゴリズムを提案する。提案手法は、データ数がNのとき、O(√kN)の計算量で解を求めることができる。同じオーダーの計算量を持つ既存の量子アルゴリズムもあるが、提案手法は近似計算に適用できる可能性を持つ。"}},{"key":"Utsumi2020_PRMU","id":"1805","type":"other","language":"JA","show":"show","authors":[{"ja":"内海 ゆづ子","en":""},{"ja":"岩口 尭史","en":""},{"ja":"汪 雪婷","en":""},{"ja":"菅沼 雅徳","en":""},{"ja":"西村 真衣","en":""},{"ja":"原 健翔","en":""},{"ja":"平川 翼","en":""},{"ja":"福井 宏","en":""}],"title":{"ja":"MIRU2020若手プログラム実施報告"},"journal":{"ja":"電子情報通信学会技術報告"},"volume":"120","number":"300","presen_id":"PRMU2020-53","pages":"86--92","year":2020,"month":12,"review":"not_reviewed","fields":[]},{"key":"Yamada2019_IEEE","id":"1708","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Yoshihiro Yamada","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Takuya Akiba","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"ShakeDrop Regularization for Deep Residual Learning"},"journal":{"ja":"IEEE Access"},"volume":"7","number":"1","pages":"186126-186136","year":2019,"month":12,"review":"reviewed","fields":["deep-learning"],"doi":"10.1109/ACCESS.2019.2960566","abstract":{"ja":"Overfitting is a crucial problem in deep neural networks, even in the latest network architectures. In this paper, to relieve the overfitting effect of ResNet and its improvements (i.e., Wide ResNet, PyramidNet, and ResNeXt), we propose a new regularization method called ShakeDrop regularization. ShakeDrop is inspired by Shake-Shake, which is an effective regularization method, but can be applied to ResNeXt only. ShakeDrop is more effective than Shake-Shake and can be applied not only to ResNeXt but also ResNet, Wide ResNet, and PyramidNet. An important key is to achieve stability of training. Because effective regularization often causes unstable training, we introduce a training stabilizer, which is an unusual use of an existing regularizer. Through experiments under various conditions, we demonstrate the conditions under which ShakeDrop works well."}},{"key":"Augereau2019","id":"1802","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Olivier Augereau","en":""},{"ja":"Kai Kunze","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Experimental Supplements from Mobile Tools for Cognitive Introspection Towards Cognitive Augmentation."},"journal":{"ja":"GetMobile Mob. Comput. Commun.","en":"GetMobile Mob. Comput. Commun."},"volume":"23","number":"2","pages":"22-24","year":2019,"month":11,"review":"reviewed","fields":[],"doi":"10.1145/3372300.3372305"},{"key":"Kudo2019_J","id":"1679","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"工藤力","en":"Riki KUDO"},{"ja":"岩田基","en":"Motoi IWATA"},{"ja":"岩村雅一","en":"Masakazu IWAMURA"},{"ja":"黄瀬浩一","en":"Koichi KISE"}],"title":{"ja":"自動付与したウィーン分類コードと図形の形状に基づく類似商標検索法","en":"A Method of Trademark Search Based on the Automatically Assigned Codes of Vienna Classification and Figure Shapes"},"journal":{"ja":"日本画像学会誌","en":"Journal of the Imaging Society of Japan"},"volume":"58","number":"3","pages":"274-283","year":2019,"month":6,"review":"reviewed","fields":["information-retrieval","document-image-analysis","pattern-recognition"],"doi":"10.11370/isj.58.274","url":"https://www.jstage.jst.go.jp/article/isj/58/3/58_274/_article/-char/en","abstract":{"ja":"年々増加する商標の登録作業を効率化するため，類似商標検索が求められている．従来の検索手法は，外観の類似性を図形の形状の類似性からのみ評価するものが大半である．ところが，商標審査官の類似性判断では，形状の類似性だけではなく，商標に描かれている図形要素の分類も加味されることが多い．本研究では，この機能の実現のため，商標の図形分類指標であるウィーン分類に着目する．2枚の商標に同じウィーン分類コードが付いていれば，それらは同じ分類の図形を含むため，類似している可能性が高い．つまりウィーン分類コードの照合で，図形の分類を同じくする類似商標を検索できると言える．この場合，検索質問の商標へのウィーン分類コードの付与が必須である．現在ウィーン分類コードの付与は，専門家が知識と経験則をもって行っている．そのスキルをDeep Learningを用いて学習することで，我々は自動でウィーン分類コードを付与し，検索できる手法を提案する．本手法は，従来のウィーン分類を用いない手法と比べて3ポイント検索精度が向上した．","en":"Demand for the search of similar trademarks is increasing. In the existing methods, the similarity is measured based on the similarity of appearance with the shape. However, to judge the similarity of appearance, trademark examiners also take into account the classification of graphical elements in trademarks. To realize this function, we focus on the Vienna Classification which is the classification of graphical elements in trademarks. If two trademarks have the same Vienna Classification code, they tend to have similar contents. However, the codes are currently assigned by experts, which prevent us from using the codes in search methods. We solve this problem by automatic assignment of Vienna Classification codes implemented using the Deep Learning. The proposed method takes into account both the automatically assigned codes and shape similarity for search. We have succeeded to improve a three percentage point accuracy as compared with the method without using the Vienna Classification."}},{"key":"Yuasa2019","id":"1706","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Shigeaki Yuasa","en":""},{"ja":"Takumi Nakai","en":""},{"ja":"Takanori Maruichi","en":""},{"ja":"Manuel Landsmann","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Masaki Matsubara","en":""},{"ja":"Atsuyuki Morishima","en":""}],"title":{"ja":"Towards Quality Assessment of Crowdworker Output Based on Behavioral Data"},"journal":{"ja":"Proc. of The Third Human-in-the-loop Methods and Human-Machine Collaboration in BigData, IEEE Big Data"},"num_pages":"3","year":2019,"month":12,"review":"reviewed","presen_type":"oral","fields":["crowdsourcing","activity-recognition","sensing"],"abstract":{"ja":"In this paper, we show preliminary results on the quality assessment of crowdworker output based on the movements of the mouse and the eyes while the task is performed. We assume that the mouse and the eyes stop longer if the quality is lower due to the lack of knowledge, or confidence, etc. Because the mouse- and eye-stopping duration follows log-normal distribution, we estimate its parameters (mean and standard deviation) to evaluate the quality. Results of preliminary experiments with 10 participants show that the parameters of correct outputs are different from those of incorrect ones. As compared to the task duration, which is often used as a feature for assessment, we have found that the mouse- and the eye-stopping duration is advantageous and complementary for the assessment."}},{"key":"Ragozin2019","id":"1698","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Kirill Ragozin","en":""},{"ja":"Yun Suen Pai","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Hochen Kerdels","en":""},{"ja":"Kai Kunze","en":""}],"title":{"ja":"Private Reader: Using Eye Tracking to Improve Reading Privacy in Public Spaces"},"journal":{"ja":"Proceedings of the 21st International Conference on Human-Computer Interaction with Mobile Devices and Services"},"num_pages":"6","year":2019,"month":10,"fields":["reading-life-log"],"doi":"10.1145/3338286.3340129","abstract":{"ja":"Reading in public spaces can often be tricky if we wish to keep the contents away from the prying eye. We propose Private Reader, an eye-tracking approach towards maintaining privacy while reading by rendering only the portion of text that is gazed by the reader. We conducted a user study by evaluating for both the reader and observer in terms of privacy, reading comfort, and reading speed for three reading modes; normal, underscored, and scrambled text. \"Scrambled\" performs best in terms of perceived effort and frustration for the shoulder surfer. Our contribution is threefold; we developed a system to preserve privacy by rendering only the text at gaze-point of the reader, we conducted a user study to evaluate user preferences and subjective task load, and we suggested several scenarios where Private Reader is useful in public spaces."}},{"key":"Ishimaru2019","id":"1693","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Shoya Ishimaru","en":"Shoya Ishimaru"},{"ja":"Takanori Maruichi","en":"Takanori Maruichi"},{"ja":"Manuel Landsmann","en":"Manuel Landsmann"},{"ja":"Koichi Kise","en":"Koichi Kise"},{"ja":"Andreas Dengel","en":"Andreas Dengel"}],"title":{"ja":"Electrooculography dataset for reading detection in the wild","en":"Electrooculography dataset for reading detection in the wild"},"journal":{"ja":"Adjunct Proceedings of the 2019 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of the 2019 ACM International Symposium on Wearable Computers","en":"Adjunct Proceedings of the 2019 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of the 2019 ACM International Symposium on Wearable Computers"},"pages":"85-88","year":2019,"month":9,"review":"not_reviewed","fields":["reading-life-log"]},{"key":"Maruichi2019_NHDI","id":"1701","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Takanori Maruichi","en":""},{"ja":"Shoya Ishimaru","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Self-confidence Estimation on Vocabulary Tests with Stroke-level Handwriting Logs"},"journal":{"en":"Proceedings of 2nd International Workshop on Human-Document Interaction"},"num_pages":"5","year":2019,"month":9,"review":"reviewed","presen_type":"oral","fields":["learning-analytics","sensing"],"abstract":{"ja":"A user’s self-confidence is as important as the correctness of the answer. Giving feedback based on self-confidence will enable the user to review unacquired knowledge more efficiently. In this paper, we propose a self-confidence estimation method based on stroke-level handwriting behavior. We consider two different handwriting answer formats: the format with character writing box and the format without it (free format). We have conducted the experiments to compare the performance on the dataset recorded with these two formats. We successfully built a method that can estimate the user’s self-confidence with maximum 80% accuracy in the dataset with character writing box, 79% in the dataset in the free format."}},{"key":"Kazuaki2019_ACM","id":"1684","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Nomura Kazuaki","en":"Nomura Kazuaki"},{"ja":"Motoi Iwata","en":"Motoi Iwata"},{"ja":"Olivier Augereau","en":"Olivier Augereau"},{"ja":"Koichi Kise","en":"Koichi Kise"}],"title":{"ja":"Estimation of Student's Engagement Based on the Posture","en":"Estimation of Student's Engagement Based on the Posture"},"journal":{"ja":"Proceedings of the 2019 ACM International Joint Conference and 2019  International Symposium on Pervasive and Ubiquitous Computing and Wearable Computers"},"num_pages":"4","year":2019,"month":9,"location":"UbiComp2019","review":"reviewed","presen_type":"poster","fields":["learning-analytics"],"doi":"10.1145/3341162.3343767","url":"https://doi.org/10.1145/3341162.3343767"},{"key":"Landsmann2019_ACM","id":"1690","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Manuel Landsmann","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Classification of Reading and Not Reading Behavior Based on Eye Movement Analysis"},"journal":{"ja":"Proceedings of the 2019 ACM International Joint Conference and 2019  International Symposium on Pervasive and Ubiquitous Computing and Wearable Computers"},"num_pages":"4","year":2019,"month":9,"location":"UbiComp2019","review":"reviewed","presen_type":"poster","fields":["learning-analytics"],"doi":"10.1145/3341162.3343811","url":"https://doi.org/10.1145/3341162.3343811","abstract":{"ja":"Nowadays, many researchers analyze reading behavior with eye trackers. Various traits of reading like engagement, or text difficulty have been observed in laboratory settings. But,\ntheir automatic application for daily life is usually prevented by one question: when is somebody reading? We have developed a tool to classify short sequences of fixations from eye gaze data into reading and not reading. Our specific use case is the Vocabulometer, a website for learning English by reading texts. We used supervised learning on data from nonnative English speakers to train decision trees for the\nclassification. With features based on vertical eye movement, we achieved 93.1% of correct classifications."}},{"key":"Miyamoto2019_AQISa","id":"1680","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masayuki Miyamoto","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"A Quantum Algorithm for Minimum Steiner Tree Problem"},"journal":{"ja":"Proc. 19th Asian Quantum Information Science Conference (AQIS2019)"},"num_pages":"2","year":2019,"month":8,"location":"Seoul, Korea","review":"reviewed","presen_type":"poster","fields":["quantum-computing"],"note":{"ja":"Full version of the paper is available at <a href=\"https://arxiv.org/abs/1904.03581\">https://arxiv.org/abs/1904.03581</a>"},"abstract":{"ja":"Minimum Steiner tree problem is a well-known NP-hard problem. \nFor the minimum Steiner tree problem in graphs with $n$ vertices and $k$ terminals, there are many classical algorithms that take exponential time in $k$. \nIn this paper, to the best of our knowledge, we propose the first quantum algorithm for the minimum Steiner tree problem.\nThe complexity of our algorithm is $\\mathcal{O}^*(1.812^k)$. \nA key to realize the proposed method is how to reduce the computational time of dynamic programming by using a quantum algorithm because existing classical (non-quantum) algorithms in the problem rely on dynamic programming. Fortunately, dynamic programming is realized by a quantum algorithm for the travelling salesman problem, in which Grover's quantum search algorithm is introduced. However, due to difference between their problem and our problem to be solved, recursions are different. Hence, we cannot apply their technique to the minimum Steiner tree problem in that shape. We solve this issue by introducing a decomposition of a graph proposed by Fuchs et al."}},{"key":"Miyamoto2019_AQISb","id":"1681","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Kohei Miyamoto","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"A Quantum Algorithm for Finding k-Minima"},"journal":{"ja":"Proc. 19th Asian Quantum Information Science Conference (AQIS2019)"},"num_pages":"4","year":2019,"month":8,"location":"Seoul, Korea","review":"reviewed","presen_type":"poster","fields":["quantum-computing","nearest-neighbor-search"],"note":{"ja":"Full version of the paper is available at <a href=\"https://arxiv.org/abs/1907.03315\">https://arxiv.org/abs/1907.03315</a>"},"abstract":{"ja":"We propose a new \\textit{finding $k$-minima} algorithm and prove that the query complexity is $\\mathcal{O}(\\sqrt{kN})$, where $N$ is the number of data indices.\nThe primary difficulty of the problem is that it requires to return $k$ answers.\nFor the problem, an extension of the Amplitude Amplification (we call it \\textit{searching all marked $k$ indices} algorithm) finds all $k$ data with the query complexity of $\\mathcal{O}(\\sqrt{kN})$ if an appropriate threshold is given.\nWe give a quantum algorithm that searches a good threshold with the complexity of $\\mathcal{O}(\\sqrt{N}\\log{k})$.\nIn addition, we briefly prove the query complexity of the \\textit{searching all marked $k$-indices} algorithm, which is not well discussed so far.\nOur algorithm can be directly adapted to distance-related problems like $k$-Nearest Neighbor Search, clustering and classification."}},{"key":"Utsumi2019_CVPPP","id":"1678","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"","en":"Yuzuko Utsumi"},{"ja":"","en":"Koichiro Nakamura"},{"ja":"","en":"Masakazu Iwamura"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"DNN-Based Tiller Number Estimation for Coping with Shortage of Labeled Data"},"journal":{"en":"CVPR 2019 workshop on Computer Vision Problems in Plant Phenotyping (CVPPP 2019)"},"presen_id":"27","num_pages":"1","year":2019,"month":6,"location":"Long Beach, CA, USA","review":"reviewed","fields":["plant-image-processing","pattern-recognition","deep-learning"],"url":"https://www.plant-phenotyping.org/lw_resource/datapool/systemfiles/elements/files/42e1416f-70a4-11e9-b1c5-dead53a91d31/current/document/UtsumiCVPPP2019.pdf"},{"key":"Sakamoto2019","id":"1710","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"坂本周司","en":""},{"ja":"山田良博","en":""},{"ja":"岩村雅一","en":""},{"ja":"岩田基","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"読書行動識別のためのSelf-supervised Learning の実験的検討"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"119","number":"348","series":"HIP2019-71","pages":"35-40","num_pages":"6","year":2019,"month":12,"location":"東北大学","review":"not_reviewed","fields":["sensing","deep-learning"]},{"key":"Daiku2019","id":"1689","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"大工 勇樹","en":""},{"ja":"岩田 基","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"英語多読学習支援のための Manga Vocabulometer の開発"},"journal":{"ja":"電子情報通信学会 ヒューマンコミュニケーショングループ 第二種時限研究会 第2回コミック工学研究会"},"pages":"19-24","num_pages":"6","year":2019,"month":10,"location":"北海道函館市","review":"not_reviewed","presen_type":"oral","fields":["sensing","learning-analytics"]},{"key":"Imagawa2019","id":"1703","type":"domestic_conf","language":"JA","show":"show_except_abstract","authors":[{"ja":"今川 寛之","en":"Hiroyuki IMAGAWA"},{"ja":"岩田 基","en":"Motoi IWATA"},{"ja":"黄瀬 浩一","en":"Koichi KISE"}],"title":{"ja":"位置検出に深層学習を利用した印刷物用電子透かし法","en":"Digital watermarking method against print-cam attack using deep learning for synchronization"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"119","number":"207","presen_id":"EMM2019-69","pages":"31-36","num_pages":"6","year":2019,"month":9,"review":"not_reviewed","fields":[]},{"key":"Landsmann2019","id":"1691","type":"domestic_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Manuel Landsmann","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Detecting reading based on eye movement analysis"},"num_pages":"6","year":2019,"month":7,"location":"人工知能学会 先進的学習科学と工学研究会(ALST)","review":"not_reviewed","fields":["learning-analytics"],"abstract":{"ja":"As, nowadays, eye trackers have become more accurate and affordable, their use also increased. One of their applied fields is the analysis of reading behavior. In laboratory settings, various traits of reading like engagement, or text difficulty could be shown. But, there is usually one unanswered question preventing their automatic application in daily life: when is somebody reading? We have developed a tool that uses eye gaze data to detect the period of reading. Our specific use case is the vocabulometer, a website for learning English by reading texts. We used supervised learning on data from nonnative English speakers to train decision trees on reading detection. With features based on vertical eye movement, an accuracy of 93.5% on detecting reading and 92.7% on detecting not reading could be achieved."}},{"key":"Maruichi2019","id":"1683","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"丸市 賢功","en":"Takanori Maruichi"},{"ja":"石丸 翔也","en":"Shoya Ishimaru"},{"ja":"Augereau Olivier","en":"Olivier Augereau"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"英単語問題解答時の確信判定システムの精度検証"},"year":2019,"month":7,"location":"先進的学習科学と工学研究会","review":"not_reviewed","fields":["sensing","learning-analytics","pattern-recognition"],"abstract":{"en":"Feedback based on the combination of self-con dence and correctness of an answer canhelp learners identify misconceptions. In this study, we propose a self-con dence estimation methodto aid the development of a system for vocabulary learning. The proposed method is based onthe examination of eye movement and handwriting behaviors. To evaluate the performance of theproposed method, we examined the eye movement of 7 participants and the handwriting behaviorof 8 participants while completing English spelling tests. The results revealed that our methodwas able to classify self-con dence into two classes with 87% accuracy in the typing dataset and79% in the handwriting dataset."}},{"key":"Nomura2019","id":"1682","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"野村 和朗","en":"Kazuaki Nomura"},{"ja":"Olivier Augereau","en":"Olivier Augereau"},{"ja":"岩田 基","en":"Motoi Iwata"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"シート圧力センサと Web カメラを用いた映像授業視聴時の集中度推定","en":"Estimation of Student’s Engagement with a Pressure Mat and a Web Camera"},"year":2019,"month":7,"location":"人工知能学会 先進的学習科学と工学研究会(ALST)","review":"not_reviewed","fields":["learning-analytics","sensing"],"abstract":{"ja":"It is important for teachers to grasp students’ engagement in order to improve the quality of lectures. When they find that their students are not engaged in the lecture, they can give advice to the students to pay attention to the lecture. However, in the e-learning environment, there is no teacher to grasp the student’s engagement. So if the student loses his engagement, no one can help him to regain it. It may cause ineffective learning. The purpose of this study is to grasp the student’s engagement by using a pressure mat and web camera. We recorded the students’ postural data, that is upper body pressure distribution and upper body pose information, while they were taking e-learning lectures. In order to get the body to pose information from a web camera, we used a human pose estimation library called OpenPose. Then we extracted 38 features from upper body pressure distribution and 33 features from upper body pose information for every minute. We selected effective features by using the Forward Stepwise Selection. Lastly, we estimate whether he or she was engaged in or not with a Support Vector Machine. As a result, the average accuracy was 79.3% for student-dependent estimation. This result shows it is possible to predictthe student’s engagement automatically.","en":"It is important for teachers to grasp students’ engagement in order to improve the quality of lectures. When they find that their students are not engaged in the lecture, they can give advice to the students to pay attention to the lecture. However, in the e-learning environment, there is no teacher to grasp the student’s engagement. So if the student loses his engagement, no one can help him to regain it. It may cause ineffective learning. The purpose of this study is to grasp the student’s engagement by using a pressure mat and web camera. We recorded the students’ postural data, that is upper body pressure distribution and upper body pose information, while they were taking e-learning lectures. In order to get the body to pose information from a web camera, we used a human pose estimation library called OpenPose. Then we extracted 38 features from upper body pressure distribution and 33 features from upper body pose information for every minute. We selected effective features by using the Forward Stepwise Selection. Lastly, we estimate whether he or she was engaged in or not with a Support Vector Machine. As a result, the average accuracy was 79.3% for student-dependent estimation. This result shows it is possible to predictthe student’s engagement automatically."}},{"key":"Taniguchi2019_CVIM","id":"1673","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"谷口 善信","en":""},{"ja":"Muhammad Naseer Bajwa","en":""},{"ja":"岩村 雅一","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"黄瀬 浩一","en":""},{"ja":"Sheraz Ahmed","en":""}],"title":{"ja":"Fine-Grained Classificationによる糖尿病性網膜症の画像診断モデルの提案"},"journal":{"ja":"情報処理学会研究報告コンピュータビジョンとイメージメディア（CVIM）"},"presen_id":"17","num_pages":"2","year":2019,"month":6,"location":"国立オリンピック記念青少年総合センター","review":"not_reviewed","fields":["deep-learning","pattern-recognition"]},{"key":"Kawai2019_CVIM","id":"1670","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"河合隆哲","en":""},{"ja":"岩村雅一","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"視覚障碍者のための周辺情報選択システム"},"journal":{"ja":"情報処理学会研究報告コンピュータビジョンとイメージメディア（CVIM）"},"presen_id":"15","num_pages":"6","year":2019,"month":5,"location":"国立オリンピック記念青少年総合センター","review":"not_reviewed","fields":["visually-impaired-assistance"]},{"key":"Muta2019_CVIM","id":"1671","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"牟田 薫","en":""},{"ja":"Muhammad Naseer Bajwa","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"Sheraz Ahmed","en":""}],"title":{"ja":"CNNを用いた皮膚病の画像分類手法の検討〜皮膚病に対する自動診断システムの構築に向けて〜"},"journal":{"ja":"情報処理学会研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"CVIM2019","number":"18","pages":"1-2","num_pages":"2","year":2019,"month":5,"location":"国立オリンピック記念青少年総合センター","review":"not_reviewed","fields":["deep-learning","object-recognition"]},{"key":"Nakamura2019_Conf","id":"1675","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"中村浩一朗","en":""},{"ja":"内海ゆづ子","en":""},{"ja":"岩村雅一","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"小規模な正解ラベル付きデータを用いたCNN に基づくエノコログサの分げつ数の推定"},"journal":{"ja":"農業情報学会2019年度年次大会公演要旨集"},"presen_id":"個-02","pages":"53-54","num_pages":"2","year":2019,"month":5,"fields":["plant-image-processing","pattern-recognition","deep-learning"]},{"key":"Utsumi2019_PRMU","id":"1676","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"内海ゆづ子","en":""},{"ja":"中村浩一朗","en":""},{"ja":"岩村雅一","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"Pretext taskを用いた植物画像からの分げつ数の推定"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"119","number":"64","presen_id":"PRMU2019-8,","pages":"265-270","num_pages":"6","year":2019,"month":5,"review":"not_reviewed","fields":["plant-image-processing","deep-learning","pattern-recognition"],"abstract":{"ja":"農学や生物学において，植物の生育状態の把握や植物の特徴を知るため，植物の形態形質（形状やサイズ\nなどの見た目の特徴）の計測が必要不可欠である．しかし，形質によっては計測に非常に手間がかかることから，計\n測の自動化が求められている．本稿では，植物の形質の中でも，イネなどの単子葉植物の枝分かれの様子を表す分げ\nつに焦点をあて，画像から分げつ数を自動的に推定する手法の提案をする．現在，画像認識の分野では，Deep Neural\nNetwork (DNN) に基づく手法が圧倒的な性能を発揮しており，植物計測に対してもDNN の利用が有効であると考え\nられる．しかし，DNN では大量の学習データが必要とされるため，分げつのような計測に手間がかかるものに対し\nて，大量の正解ラベル付き学習データを用意することが難しい．そこで，本研究では，分げつ数が不明な大量の植物\n画像を用いて，自動的に正解ラベルを付与できるpretext task によりネットワークを学習し，これを分げつ数の推定\nに転用する．このことにより，小規模なラベル付きデータしか用意できない分げつ数の推定に対して，DNN を適用可\n能となる．実験の結果，pretext task を用いた一部の推定器が，従来の計測値に基づいて構築した分げつ数推定モデ\nルよりもよりよい推定結果を示すことが明らかとなった．"}},{"key":"Iwamura2019","id":"1641","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":""},{"ja":"南谷 和範","en":""},{"ja":"井上 慶彦","en":""},{"ja":"程 征","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"全方位カメラと認識技術を用いた視覚障碍者支援システム"},"journal":{"ja":"情報アクセシビリティをめぐる諸問題に関する研究集会"},"year":2019,"month":3,"review":"not_reviewed","presen_type":"oral","fields":["visually-impaired-assistance"]},{"key":"Yamada2019_PRMU","id":"1651","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"山田良博","en":""},{"ja":"岩村雅一","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"画像分類精度を向上させる追加タスクの検討"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"118","number":"513","presen_id":"PRMU2018-159","pages":"155-160","num_pages":"6","year":2019,"month":3,"location":"電気通信大学, 東京都調布市","fields":["object-recognition","deep-learning"],"abstract":{"ja":"画像中に含まれる物体を分類する画像分類技術において，Convolutional Neural Network (CNN)は分類精度を向上させることで注目を集めている．原則として，CNNの学習は各画像に正しい分類結果を表すラベルが付与されているという条件が仮定されている．この場合，優れた特徴表現能力を学習するためには，人手でラベルが付けられた膨大な枚数の画像を学習する必要があり，コストが高い．この状況を改善するために，ラベルが付与されていない条件で優れた特徴表現能力を獲得する手法が模索されていた．近年，CNNに対して本来の画像分類とは異なる学習タスク(Pretext Task)を扱うことで，ラベルが付与されていない条件で優れた特徴表現が学習可能なことが示された．このようなPretext Taskは従来の教師あり学習とは異なる特徴表現能力を学習している可能性が高く，組み合わせることで分類精度を改善できると考えられる．しかし，Pretext Taskと本来の画像分類タスクを組み合わせた場合については詳しく検証されておらず，有効な方法は示されていない．そこで，2種類のアプローチを通して，Pretext Taskと本来の画像分類タスクについての有効性について検討を行った．実験結果からは，全くランダムに与えられたラベルの学習であっても，本来の画像分類タスクに寄与することを確認した．ただし，我々が検討したタスクは認識精度を大きく改善しなかった．"}},{"key":"Sonoda2019_PRMU","id":"1649","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"薗田 統輔","en":"Tosuke SONODA"},{"ja":"岩田 基","en":"Motoi IWATA"},{"ja":"稲富 宏之","en":"Hiroyuki INADOMI"},{"ja":"黄瀬浩一","en":"Koichi KISE"}],"title":{"ja":"Convolutional Neural Network を用いた樹木画の画型分類の改善","en":"Improvement of Tree-Drawing Classification Using Convolutional Neural Network"},"journal":{"ja":"電子情報通信学会技術研究報告パターン認識メディア理解研究会"},"presen_id":"PRMU2018-158","pages":"149-154","num_pages":"6","year":2019,"month":3,"location":"電気通信大学","review":"not_reviewed","fields":["deep-learning"]},{"key":"Kudo2019_PhD","id":"1640","type":"doctoral_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"工藤 力","en":""}],"title":{"ja":"図形の形状と分類に着目した類似商標検索に関する研究"},"year":2019,"month":2,"review":"not_reviewed","fields":[]},{"key":"Sanches2019","id":"1791","type":"doctoral_thesis","language":"EN","show":"show_except_abstract","authors":[{"ja":"Charles Lima Sanches","en":"Charles Lima Sanches"}],"title":{"ja":"Reading behavior analysis via non-Professional eye trackers to enhance knowledge acquisition","en":"Reading behavior analysis via non-Professional eye trackers to enhance knowledge acquisition"},"year":2019,"month":2,"review":"not_reviewed","fields":[]},{"key":"Utsumi2019","id":"1759","type":"survey","language":"JA","show":"show","authors":[{"ja":"内海 ゆづ子","en":""}],"title":{"ja":"画像処理・認識技術を用いた植物表現型計測"},"journal":{"ja":"Medical Imaging Technology"},"volume":"37","number":"4","series":"September","pages":"172-176","year":2019,"month":9,"fields":["plant-image-processing","pattern-recognition"]},{"key":"Iwata2019_IL_Howb","id":"1749","type":"survey","language":"JA","show":"show_except_datafile","authors":[{"ja":"岩田 基","en":"Motoi Iwata"},{"ja":"Olivier Augereau","en":"Olivier Augereau"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"視線情報に基づいたスマートフォン上での漫画表示手法","en":"How to display comics on smartphones based on eye tracking"},"journal":{"ja":"画像ラボ","en":"Image Laboratory"},"volume":"30","number":"8","pages":"21-26","num_pages":"6","year":2019,"month":9,"publisher":"日本工業出版","review":"not_reviewed","fields":["comic-computing","sensing"]},{"key":"Utsumi2019_MIT_Phenotypinga","id":"1687","type":"other","language":"JA","show":"show","authors":[{"ja":"内海 ゆづ子","en":"Yuzuko Utsumi"}],"title":{"ja":"画像処理・認識技術を用いた植物表現型計測","en":"Phenotyping Using Image Processing and Pattern Recognition Technology"},"journal":{"ja":"Medical Imaging Technology","en":"Medical Imaging Technology"},"volume":"37","number":"4","pages":"172--176","num_pages":"6","year":2019,"month":10,"fields":["plant-image-processing","pattern-recognition"],"doi":"10.11409/mit.37.172","url":"https://www.jstage.jst.go.jp/article/mit/37/4/37_172/_article/-char/ja/"},{"key":"Utsumi2019_MIT_Phenotypingb","id":"1688","type":"other","language":"JA","show":"show","authors":[{"ja":"内海 ゆづ子","en":"Yuzuko Utsumi"}],"title":{"ja":"画像処理・認識技術を用いた植物表現型計測","en":"Phenotyping Using Image Processing and Pattern Recognition Technology"},"journal":{"ja":"Medical Imaging Technology","en":"Medical Imaging Technology"},"volume":"37","number":"4","pages":"172--176","num_pages":"6","year":2019,"month":10,"fields":["plant-image-processing"],"doi":"10.11409/mit.37.172","url":"https://www.jstage.jst.go.jp/article/mit/37/4/37_172/_article/-char/ja/"},{"key":"Utsumi2019_MIT","id":"1711","type":"other","language":"JA","show":"show","authors":[{"ja":"内海 ゆづ子","en":"Yuzuko Utsumi"}],"title":{"ja":"画像処理・認識技術を用いた植物表現型計測"},"journal":{"ja":"Medical Imaging Technology","en":"Medical Imaging Technology"},"volume":"37","number":"4","pages":"172--176","year":2019,"month":9,"fields":["plant-image-processing","pattern-recognition"],"doi":"10.11409/mit.37.172","url":"https://www.jstage.jst.go.jp/article/mit/37/4/37_172/_article/-char/ja","abstract":{"ja":"農業の分野では，育種（品種改良）に遺伝子情報が用いられている．育種に遺伝子情報を利用するには，遺伝子と表現型の対応を調査する必要がある．DNAシークエンサーの発達により遺伝子の読み取りは容易に行えるようになってきたものの，その遺伝子が植物の形状や生育にどのように影響を与えているかはいまだ不明なことが多い．そこで，遺伝子の異なる植物を大量に生育し，表現型を計測するフェノタイピングが盛んに行われるようになってきた．画像処理・認識技術は，非破壊の計測が可能であること，形状だけでなく色情報も同時に計測できることから，現在植物計測の中心的な技術となっている．本稿では，画像での植物計測特有の問題点を示しながら，フェノタイピングをおもな目的とした植物の表現型計測手法を紹介する．"}},{"key":"Sanches2018","id":"1636","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Charles Lima Sanches","en":"Charles Lima Sanches"},{"ja":"Olivier Augereau","en":"Olivier Augereau"},{"ja":"Koichi Kise","en":"Koichi Kise"}],"title":{"ja":"Estimation of reading subjective understanding based on eye gaze analysis","en":"Estimation of reading subjective understanding based on eye gaze analysis"},"journal":{"ja":"PloS one","en":"PloS one"},"volume":"13","number":"10","year":2018,"month":10,"review":"reviewed","fields":["reading-life-log","sensing"],"doi":"10.1371/journal.pone.0206213","url":"https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0206213&type=printable","abstract":{"ja":"The integration of ubiquitous technologies in the field of education has considerably enhanced our way of learning.\nSuch technologies enable students to get a gradual feedback about their performance and to provide adapted learning materials.\nIt is particularly important in the domain of foreign language learning which requires intense daily practice.\nOne of the main inputs of adaptive learning systems is the user's understanding of a reading material.\nThe reader's understanding can be divided into two parts: the objective understanding and the subjective understanding.\nThe objective understanding can be measured by comprehension questions about the content of the text.\nThe subjective understanding is the reader's perception of his own understanding.\nThe subjective understanding plays an important role in the reader's motivation, self-esteem and confidence.\nHowever, its automatic estimation remains a challenging task.\nThis paper is one of the first to propose a method to estimate the subjective understanding.\nWe show that using the eye gaze to predict the subjective understanding improves the estimation by 13% as compared to using comprehension questions.","en":"The integration of ubiquitous technologies in the field of education has considerably enhanced our way of learning.\nSuch technologies enable students to get a gradual feedback about their performance and to provide adapted learning materials.\nIt is particularly important in the domain of foreign language learning which requires intense daily practice.\nOne of the main inputs of adaptive learning systems is the user's understanding of a reading material.\nThe reader's understanding can be divided into two parts: the objective understanding and the subjective understanding.\nThe objective understanding can be measured by comprehension questions about the content of the text.\nThe subjective understanding is the reader's perception of his own understanding.\nThe subjective understanding plays an important role in the reader's motivation, self-esteem and confidence.\nHowever, its automatic estimation remains a challenging task.\nThis paper is one of the first to propose a method to estimate the subjective understanding.\nWe show that using the eye gaze to predict the subjective understanding improves the estimation by 13% as compared to using comprehension questions."}},{"key":"Kudo2018_J_Proposal","id":"1621","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"工藤 力","en":"Riki Kudo"},{"ja":"岩田 基","en":"Motoi Iwata"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"高速・高精度を実現した類似商標画像検索システムの提案","en":"Proposal of Similar Image Retrieval System Using Large Scale Trademark Image Database"},"journal":{"ja":"日本画像学会誌"},"volume":"57","number":"4","pages":"403-412","num_pages":"10","year":2018,"month":8,"review":"reviewed","fields":[],"doi":"10.11370/isj.57.403","abstract":{"ja":"商標審査とは、外観や概念などの観点から、申請商標と類似した商標が登録商標内にないかを確認する作業である。現状では、人がキーワード検索を用いて、比較する登録商標を絞り込み、目視により類似性を評価しているが、労力がかかる上に、類似商標の検索漏れが避けられないという問題がある。特に、様々な類似性の中でも外観の類似は最も基本的で多数の事例を有するため、これを対象として類似商標を自動的に検索できれば、上記問題を解決する大きな一歩となる。このとき最も意識すべき点は，検索漏れを少なくした上で，いかに非類似商標の出力を抑えるかという点にある。また同時に検索にかかる計算量も少なくしなければ、実用に適さない。本論文では、特定物体認識で用いられる画像照合の手法を導入してこの問題の解決を図る。具体的には、局所特徴を用いた画像照合、局所特徴の配置などの情報を用いた検証処理の2処理である。商標検索実験を行った結果、recall 100%, precision 62.5%, 処理時間8.2秒という結果を得、商標審査官の労力を軽減できることが示された。","en":"The trademark examination is a task of confirming whether there is a similar trademark within the registered trademarks. In this task, searchers narrow down the similar trademarks to be compared by using keyword search, and they make judgement visually. However, this process is labor intensive and it is unavoidable to miss some similar trademarks. In particular, visual similarity is the most fundamental which covers many cases. Thus, we consider that solving the above problem with respect to visual similarity is an important step towards full automation. The most important point is to reduce output of dissimilar trademarks while keeping similar trademarks in the output. In addition, the computational cost must be small enough. In this paper, we focus on image matching methods used for specific object recognition to fulfill the requirements. From the experimental evaluation using  images and queries, we have obtained recall 100%, precision 62.5%, processing time 8.2[sec]. This indicates that the proposed method can reduce the workload of trademark examiners."}},{"key":"Augereau2018a","id":"1638","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Olivier Augereau","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"A Survey of Comics Research in Computer Science"},"journal":{"ja":"Journal of Imaging"},"volume":"4","number":"7","pages":"1-19","num_pages":"19","year":2018,"month":6,"publisher":"MDPI","review":"reviewed","fields":["document-image-analysis"],"doi":"10.3390/jimaging4070087","url":"http://www.mdpi.com/2313-433X/4/7/87","abstract":{"ja":"Graphic novels such as comic books and mangas are well known all over the world. The digital transition started to change the way people are reading comics: more and more on smartphones and tablets, and less and less on paper. In recent years, a wide variety of research about comics has been proposed and might change the way comics are created, distributed and read in the future. Early work focuses on low level document image analysis. Comic books are complex; they contains text, drawings, balloons, panels, onomatopoeia, etc. Different fields of computer science covered research about user interaction and content generation such as multimedia, artificial intelligence, human-computer interaction, etc. with different sets of values. We review the previous research about comics in computer science to state what has been done and give some insights about the main outlooks."}},{"key":"Kudo2018_J_Similar","id":"1627","type":"letter","language":"JA","show":"show","authors":[{"ja":"工藤 力","en":"Riki Kudo"},{"ja":"岩田 基","en":"Motoi Iwata"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"ウィーン分類と図形の類似性を考慮した類似商標検索法","en":"Similar Trademark Searching Method Considering Similarity Between Vienna Classification and Figures"},"journal":{"ja":"日本画像学会誌"},"volume":"57","number":"5","pages":"533-536","num_pages":"4","year":2018,"month":10,"review":"reviewed","fields":[],"doi":"10.11370/isj.57.533","abstract":{"ja":"近年、商標審査の効率化のために、申請商標を検索クエリとして用い、外観が類似している商標を検索するシステムが求められている。従来の我々の手法は、図形の形状の類似度を算出し、ランキング方式で類似商標を検索するものである。しかし外観の類似は、図形の形状の類似のみで判断されるわけではなく、図形の概念的な要素が同一であれば類似しているとみなされるケースがある。よって、外観の類似性を扱う上で、図形の概念的な分類も考慮する必要がある。そこで本研究では、商標に含まれる図形の概念的な分類指標であるウィーン分類に着目する。ウィーン分類の一致は、図形の概念的な分類が同一であることを表す。また、ウィーン分類は、商標に複数付与されており、各々独立した図形の分類を表現しているため、その一致数が多ければ類似性が高いと考えられる。よって、上記システムにウィーン分類の一致数を類似指標として組み込む。本論文では、ウィーン分類を考慮した検索手法の有効性を検証し、精度を約2％向上することに成功した。","en":"To improve the efficiency of trademark examination, there is a demand for a system that searches trademarks with similar appearance. We have made such a system focusing on trademark shapes. However, appearance similarity is judged not only by similarity of graphical shapes, but also by conceptual elements of graphics. So, it is necessary to take them into account.\nFor this purpose, we focus on the Vienna classification. The matching of the Vienna classification indicates that trademarks share conceptual elements. One trademark can have several conceptual elements of Vienna classification, which are independent. Thus, it is possible to consider that similarity is higher if the number of shared conceptual elements is larger. By incorporating the number of shared conceptual elements into the above system, we have succeeded to improve the accuracy by 2%."}},{"key":"Kudo2018","id":"1637","type":"intl_conf","language":"JA","show":"show","authors":[{"ja":"Riki Kudo","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Improvement of Trademark Retrieval Using Pseudo Relevance Feedback"},"journal":{"ja":"Proc. 11th Indian Conference on Computer Vision, Graphics and Image Processing"},"num_pages":"8","year":2018,"month":12,"location":"Hyderabad, India","review":"reviewed","fields":[],"doi":"10.1145/3293353.3293429"},{"key":"Augereau2018b","id":"1639","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Olivier Augereau","en":""},{"ja":"Benjamin Tag","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Mental State Analysis on Eyewear"},"book_title":"Proceedings of the 2018 ACM International Joint Conference and 2018 International Symposium on Pervasive and Ubiquitous Computing and Wearable Computers","pages":"968-973","year":2018,"month":10,"fields":["reading-life-log","learning-analytics","sensing"],"doi":"10.1145/3267305.3274119","abstract":{"ja":"Sensors can be worn on various positions on the body such as wristband, clothes, armband, ring, etc. However the head is a strategic position as our brain, eyes, ears, nose and mouth are near. We introduce a short state of the art of the existing technologies for sensing the mental states and the possible applications on eyewear. Analyzing the physical activity has been broadly covered by sensing the physical movement of the body, but mental states need more fine analysis of physiological signals such as blood volume pressure, skin conductance, and brain activity. These signals can give information about emotions and knowledge of the users."}},{"key":"Jacquet2018","id":"1629","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Clément Jacquet","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"Nicholas Journet","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Demo: Vocabulometer, a Web Platform for Ubiquitous Language Learning"},"book_title":"Proceedings of the 2018 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of the 2018 ACM International Symposium on Wearable Computers","pages":"361--364","num_pages":"4","year":2018,"month":10,"publisher":"ACM","review":"reviewed","fields":["learning-analytics","sensing"],"doi":"10.1145/3267305.3267614","abstract":{"ja":"The Vocabulometer is a reading assistant system designed to record the reading activity of a user and to extract mutual information about the users and the read documents. \nThe Vocabulometer stands as a web platform and offers services for analyzing which word is read at which time, estimating the reading skill of the user, the level of difficulty of a document, and recommending reading materials according to the user's vocabulary.\nFor now the system is based on eye tracking but in the future we plan to make a smartphone application which do not need additional hardware."}},{"key":"Ohbayashi2018","id":"1630","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Yuya Ohbayashi","en":""},{"ja":"Shoya Ishimaru","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Investigating Gaze and Physiological Features to Estimate Comprehension on E-learning Video Lectures"},"journal":{"ja":"International Interdisciplinary Symposium on Reading Experience & Analysis of Documents (READ2018)"},"year":2018,"month":10,"review":"reviewed","fields":["learning-analytics","sensing"]},{"key":"Augereau2018_READ","id":"1631","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Olivier Augereau","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Towards experiential supplements for education"},"journal":{"ja":"Proc. International Interdisciplinary Symposium on Reading Experience & Analysis of Documents (READ2018)"},"year":2018,"month":10,"review":"reviewed","fields":["learning-analytics","sensing"]},{"key":"Maruichi2018_ACM","id":"1634","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Takanori Maruichi","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Keystroke tell you how confident you are: A new vocabulary acquisition tool"},"journal":{"ja":"Proceedings of the 2018 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of the 2018 ACM International Symposium on Wearable Computers"},"year":2018,"month":10,"publisher":"ACM","review":"reviewed","fields":["learning-analytics","sensing"]},{"key":"Nomura2018","id":"1633","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Kazuaki Nomura","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Estimation of Student’s Engagement Using a Smart Chair"},"journal":{"ja":"Proceedings of the 2018 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of the 2018 ACM International Symposium on Wearable Computers, ACM"},"year":2018,"month":10,"review":"reviewed","presen_type":"poster","fields":["learning-analytics","sensing"],"doi":"10.1145/3267305.3267611","url":"https://dl.acm.org/doi/10.1145/3267305.3267611"},{"key":"Yamada2018_ICLR","id":"1613","type":"intl_conf","language":"JA","show":"show","authors":[{"ja":"Yoshihiro Yamada","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"ShakeDrop Regularization"},"journal":{"ja":"Proc. 6th International Conference on Learning Representation (ICLR) Workshop"},"pages":"1-4","num_pages":"4","year":2018,"month":4,"location":"Vancouver, Canada","review":"reviewed","fields":["object-recognition","deep-learning"],"url":"https://openreview.net/forum?id=Bymu6tJwz","abstract":{"ja":"This paper proposes a powerful regularization method named ShakeDrop regularization. ShakeDrop is inspired by Shake-Shake regularization that decreases error rates by disturbing learning. While Shake-Shake can be applied to only ResNeXt which has multiple branches, ShakeDrop can be applied to not only ResNeXt but also ResNet, Wide ResNet and PyramidNet in a memory efficient way. Important and interesting feature of ShakeDrop is that it strongly disturbs learning by multiplying even a negative factor to the output of a convolutional layer in the forward training pass. The effectiveness of ShakeDrop is confirmed by experiments on CIFAR-10/100 and Tiny ImageNet datasets."}},{"key":"Augereau2018_IAPR","id":"1615","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Olivier Augereau","en":""},{"ja":"Clement Jacquet","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Nicholas Journet","en":""}],"title":{"ja":"Vocabulometer: An eye-tracking web platform for reading analysis"},"journal":{"ja":"Proceedings of the IAPR International Workshop on Document Analysis Systems"},"year":2018,"month":4,"review":"reviewed","fields":["document-image-analysis","reading-life-log"],"abstract":{"ja":"The eye-tracking technology reached a new\nbreakthrough since the last years with the development of\nlow-cost eye trackers. The technology is now accessible for\nmany people, which allows to use data mining and learning\nalgorithms to analyze the reading behavior of different\nreaders. However, almost no eye gaze database are available\nfor the research community. We propose a web platform\nfor large scale recordings through a web interface named\nVocabulometer. To encourage people to use this platform\nwe provide service linked to reading and learning language\nsuch as counting the number of read words, analyzing the\ncomprehension of the reading, predicting the difficult words\nor recommending text according to the reader skills. For now\nthe interface is compatible with inexpensive eye trackers such\nas Tobii Eye X and Tobii 4C. In the future, eye-tracking with\nwebcam could be supported."}},{"key":"Daiku2018_IAPR","id":"1614","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Yuki Daiku","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Comics Story Representation System Based on Genre"},"journal":{"ja":"Proceedings of the IAPR International Workshop on Document Analysis Systems"},"year":2018,"month":4,"review":"reviewed","fields":["document-image-analysis"],"doi":"10.1109/ICDAR.2017.293","url":"https://ieeexplore.ieee.org/document/8270238","abstract":{"ja":"Comics is usually classified into broad categories called “genres” according to its contents such as comedy, horror, science fiction, etc. Because a genre expresses a comics story briefly, people read comics which has contents based on their interest, by relying on comics genres. However, giving only one genre to one comic cannot express the detailed difference of the story. In this paper, we propose a system for generating comics story representation as a sub-sequence of genres. Our comics story representation can be applied to a new search engine based on stories or to a recommendation system which analyzes the tastes of the user’s favorite comics by finding comics with similar story representation. We use a deep neural network to classify each page into the corresponding genre. Experimental results confirm the advantage of the proposed system."}},{"key":"Brishtel2018","id":"1612","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"","en":"Iuliia Brishtel"},{"ja":"","en":"Shoya Ishimaru"},{"ja":"","en":"Olivier Augereau"},{"ja":"","en":"Koichi Kise"},{"ja":"","en":"Andreas Dengel"}],"title":{"en":"Assessing Cognitive Workload on Printed and Electronic Media using Eye-Tracker and EDA Wristband"},"journal":{"en":"Proceedings of the 23rd International Conference on Intelligent User Interfaces Companion"},"presen_id":"Article No.45","num_pages":"2","year":2018,"month":3,"location":"Tokyo, Japan","fields":[],"doi":"10.1145/3180308.3180354","url":"https://dl.acm.org/citation.cfm?id=3180354"},{"key":"Maeda2018","id":"1643","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"前田裕介","en":"Yusuke MAEDA"},{"ja":"岩田基","en":"Motoi IWATAy"},{"ja":"黄瀬浩一","en":"Koichi KISEy"}],"title":{"ja":"漫画の立体表示を目標としたコマ画像の奥行推定","en":"Depth Estimation of Panel Image for Stereoscopic Display of Comics"},"series":"HCGシンポジウム2018","presen_id":"B-5-3","pages":"5","year":2018,"month":12,"location":"シンフォニアテクノロジー響ホール伊勢","review":"not_reviewed","fields":["deep-learning"],"awards":{"ja":["特集テーマセッション賞（コミック工学）受賞"]},"abstract":{"ja":"現在漫画は電子端末を用いて電子書籍として読まれている。これに伴い紙媒体では行えなかった表現や利用が可能になるが漫画を読む楽しみ方は紙媒体で読むのとあまり変わっていない。そこで電子漫画に新しい表現を加えることで漫画をより面白く読むことができる。本稿では漫画を立体的に表示することを考える。漫画のコマ内の物体を3D 空間上に配置することでより臨場感を持たせることができる。そこでコマ内の物体の奥行の推定を行った。","en":"A comic is read as electronic books using electronic terminals．Along with this，expressions and use that could not be done with paper media will be possible，but how to enjoy comic is the same as reading on paper medium．So，by adding new expressions to electronic comic books，it is possible to read comic more interestingly. We think about displaying comic in three dimensions．By placing objects in comic strips in 3D space， ｗ e can make them more realistic．Then we estimated the depth of the object in the comic panel．"}},{"key":"Yamada2018_MIRU","id":"1628","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"山田 良博","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"ResNetsに対する新たな正則化手法ShakeDropの提案"},"journal":{"ja":"第21回画像の認識・理解シンポジウム(MIRU2018)予稿集"},"presen_id":"OS2-L1","num_pages":"4","year":2018,"month":8,"location":"北海道札幌市","review":"reviewed","presen_type":"oral_and_poster","fields":["deep-learning","object-recognition","pattern-recognition"],"awards":{"ja":["MIRU学生優秀賞受賞"]}},{"key":"Utsumi2018_Symp","id":"1626","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"内海ゆづ子","en":""},{"ja":"三木啓輔","en":""},{"ja":"尾形亮輔","en":""},{"ja":"大林拓実","en":""},{"ja":"三輪由佳","en":""},{"ja":"岩村雅一","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"ブドウ摘粒支援のための果房 3 次元構造推定"},"journal":{"ja":"第21回画像の認識・理解シンポジウム予稿集"},"number":"DS-7","num_pages":"4","year":2018,"month":8,"fields":["plant-image-processing"]},{"key":"Inoue2018_CVIM","id":"1619","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"井上 慶彦","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"全方位カメラを用いた物体検出とトラッキング −視覚障害者支援システムの実現に向けて−"},"journal":{"ja":"情報処理学会研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2018CVIM-212","number":"20","num_pages":"6","year":2018,"month":5,"location":"大阪大学","review":"not_reviewed","presen_type":"poster","fields":["object-recognition","visually-impaired-assistance"]},{"key":"Kawagishi2018_CVIM","id":"1618","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"川岸 樹奈","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"個人の服装を考慮した髪型画像検索手法"},"journal":{"ja":"情報処理学会研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2018CVIM-212","number":"21","pages":"1-7","year":2018,"month":5,"location":"大阪大学","review":"not_reviewed","presen_type":"poster","fields":[]},{"key":"Koboku2018_CVIM","id":"1620","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"古木 拓哉","en":""},{"ja":"岩村 雅一","en":""},{"ja":"岩田 基","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"類似意匠推薦システム実現のための意匠特徴部の検出"},"journal":{"ja":"情報処理学会研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2018CVIM-212","number":"15","pages":"1-5","year":2018,"month":5,"location":"大阪大学","review":"not_reviewed","presen_type":"poster","fields":["deep-learning"]},{"key":"Utsumi2018_Conf","id":"1617","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"内海ゆづ子","en":""},{"ja":"三木啓輔","en":""},{"ja":"尾形亮輔","en":""},{"ja":"大林拓実","en":""},{"ja":"三輪由佳","en":""},{"ja":"岩村雅一","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"画像を用いた果房の３次元構造推定に基づくブドウの摘粒支援"},"journal":{"ja":"農業情報学会2018年度年次大会"},"presen_id":"P-06","year":2018,"month":5,"review":"not_reviewed","presen_type":"poster","fields":["plant-image-processing","3d-reconstruction"],"awards":{"ja":["若手研究者イノベーション賞受賞"]},"abstract":{"ja":"生食用ブドウの栽培は，他の果樹と比較して作業時間や作業内容が多い．そのため，ブドウ栽培技術取得には経験と時間がかかり，新規参入者に大きな負荷となっている．本研究では，ブドウの栽培作業のうち，技術取得に最も経験が必要な摘粒作業の支援を目的として，果房から摘粒対象の果粒を自動的に提示する補助システムの作成を行っている．本発表では，補助システム実現に必要な画像からの果房の3次元構造の推定手法を提案する．摘粒作業では，果房の形状や果粒の密度，大きさなどを考慮しながら，果房中の果粒の数を一定に保って，整形をする．そのため，摘粒の判断をするには，果粒個数や配置，大きさなどの果房の３次元構造の情報が必要となる．提案手法では，果房の時系列画\n像から，SLAM によりカメラ位置を推定し，Instance segmentation により画像中の果粒のセグメンテーションをして，画像上での果粒の座標を求める．そして，ステレオビジョンを用いて果粒の３次元位置を復元する．実験の結果，全方位視覚センサを用いて撮影された圃場の果房の構造推定に成功した．"}},{"key":"Maruichi2018_PRMU","id":"1604","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"丸市賢功","en":"Takanori MARUICHI"},{"ja":"坂本周司","en":"Shuji SAKAMOTO"},{"ja":"OlivierAugereau","en":"Olivier AUGEREAU"},{"ja":"黄瀬浩一","en":"Koichi KISE"}],"title":{"ja":"キーロガを用いた英単語タイピングの確信判定","en":"Confidence Estimation for English Word Typing by Using a Key-Logger"},"journal":{"ja":"電子情報通信学会技術研究報告パターン認識メディア理解研究会"},"presen_id":"PRMU2017","pages":"1-6","num_pages":"6","year":2018,"month":3,"location":"青山学院大学 青山キャンパス","review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"英単語の学習において，復習は重要なプロセスである．一般的に復習は間違えた英単語に対して行われる．\nしかし，正解はしたものの正解であると確信が持てなかった単語を見落としては十分復習したとは言い切れない．そこで本研究では，英単語タイピング問題解答時の確信の有無を判定し，それを解答の正誤とともにユーザに提示することで，より効果的な復習の実現を目指す．確信の判定にはキーロガから取得したタイピングに関する情報を利用する．実験参加者12 名に英単語タイピング問題120 問に解答してもらい，その際の解答に対する確信の有無を判定した．その結果，ユーザ依存で学習を行った場合91.3%，ユーザ非依存で学習を行った場合には89.8%の平均判定精度が得られた．","en":"Reviewing plays an important role in English vocabulary acquisition. However, we tend to overlook the words that were correctly answered without confident. In this research, we estimate confidence in answering\nEnglish word typing questions to let learners reviewing more effectively. We use keystroke information acquired\nby a key-logger to estimate the confidence. In order to evaluate the proposed method, we employed the dataset\nobtained from 12 participants answering 120 questions. From the experimental results on estimating the confidence,\nthe proposed method has achieved the average accuracy of 91.3% in the case of user-dependent learning, and 89.8%\nin the case of user-independent learning."}},{"key":"Hoshika2018_PRMU","id":"1605","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"星加 健介","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"文書への依存性を考慮したTOEICスコア推定法の実験的評価"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"117","number":"442","presen_id":"PRMU2017-146","pages":"7-12","num_pages":"6","year":2018,"month":2,"location":"和歌山大学","review":"not_reviewed","fields":[]},{"key":"Raku2018","id":"1603","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"楽 卓登","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"Charles Lima Sanches","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"視点特徴とテキスト特徴の分析による日本語テキストの難易度推定","en":"Estimation of Text Difficulty by Analyzing Text and Eye Gaze Features"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"117","number":"442","series":"PRMU2017-147","pages":"pp. 13-18","num_pages":"6","year":2018,"month":2,"location":"和歌山大学","review":"not_reviewed","fields":["reading-life-log"]},{"key":"Furuya2018","id":"1602","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"古谷 舞","en":""},{"ja":"内海ゆづ子","en":""},{"ja":"松村 篤","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"Structure from motionを用いた植物根系の形態解析の検討"},"journal":{"ja":"信学技報"},"volume":"117","number":"391","presen_id":"PRMU2017-137","pages":"235-236","year":2018,"month":1,"fields":["3d-reconstruction"]},{"key":"Kometani2018","id":"1625","type":"book","language":"JA","show":"show","authors":[{"ja":"米谷竜","en":""},{"ja":"斎藤英雄 編著","en":""},{"ja":"池畑 諭","en":""},{"ja":"牛久 祥孝","en":""},{"ja":"内山 英昭","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"小野 峻佑","en":""},{"ja":"片岡 裕雄","en":""},{"ja":"金崎 朝子","en":""},{"ja":"川西 康友","en":""},{"ja":"齋藤 真樹","en":""},{"ja":"櫻田 健","en":""},{"ja":"高橋 康輔","en":""},{"ja":"松井 勇佑","en":""}],"title":{"ja":"未来へつなぐデジタルシリーズ37 コンピュータビジョン -広がる要素技術と応用-"},"year":2018,"month":6,"publisher":"共立出版","review":"not_reviewed","fields":[],"note":{"ja":"詳細は以下の通りです．\nhttp://www.kyoritsu-pub.co.jp/bookdetail/9784320123571"},"abstract":{"ja":"コンピュータによる視覚(ビジョン)機能を実現することを目的としたコンピュータビジョンは1970年代からはじまる学問分野であり，カメラで撮影された画像や映像に映る対象を理解するための様々な技術が長らく研究開発されてきた。特に近年は，深層学習をベースとした人工知能技術の発展とあいまって，基礎，応用の両面で急速な進歩を遂げている。また，仮想現実や拡張現実，マルチメディア，インタラクションなど様々な分野への応用も期待されている。\n　このような背景をのもと，本書『コンピュータビジョン―広がる要素技術と応用―』は，カメラキャリブレーションやカメラ位置姿勢推定，多視点幾何やフォトメトリックステレオによる3次元復元といった黎明期からの主要技術に加え，顔認識，人物属性認識，行動認識，重要領域検出，RGBD画像を用いた3次元物体認識，画像のキャプショニングといった応用技術，さらには近似最近傍探索，マルコフ確率場の推論と学習，凸最適化といった基盤技術をカバーする13章から構成される。本書を通して，コンピュータビジョン分野において最近注目を集めるトピックに関する基盤理論や原理，近年の重要文献，さらには実装方法を学べることが期待される。"}},{"key":"Iwamura2018","id":"1600","type":"book","language":"EN","show":"show_except_datafile","authors":[{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Advances of Scene Text Datasets"},"book_title":"Visual Text Interpretation - Algorithms and Applications in Scene Understanding and Document Analysis","year":2018,"publisher":"Springer","fields":["pattern-recognition"],"abstract":{"ja":"This article introduces publicly available datasets in scene text detection and recognition."}},{"key":"Kevin2018","id":"1784","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"モウンタニボング ケビン","en":"MOUNTHANYVONG Kevin"}],"title":{"ja":"ユーザの理解度を考慮した教材の自動選択","en":"Towards documents adapted to user's understanding"},"year":2018,"month":9,"fields":[]},{"key":"Iwata2018","id":"1635","type":"other","language":"EN","show":"show_except_datafile","authors":[{"ja":"","en":"Motoi Iwata"}],"title":{"en":"Manga as a media to convey knowledge to society"},"journal":{"en":"2018 International Conference on Documentation and Information (ICDI 2018)"},"year":2018,"month":9,"location":"Jakarta, Indonesia","review":"not_reviewed","fields":[],"url":"http://pdii.lipi.go.id/loknas/index.php/conference/icdi-pdii","note":{"en":"Invited Keynote Talk"}},{"key":"Augereau2017_PACM","id":"1591","type":"journal_paper","language":"EN","show":"show_except_datafile","authors":[{"ja":"Olivier Augereau","en":""},{"ja":"Charles Lima Sanches","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Kai Kunze","en":""}],"title":{"ja":"Wordometer Systems for Everyday Life"},"journal":{"ja":"PACM on Interactive, Mobile, Wearable and Ubiquitous Technologies"},"volume":"1","number":"4","pages":"123:1-123:21","num_pages":"21","year":2017,"month":12,"publisher":"ACM","review":"reviewed","fields":["reading-life-log"],"doi":"10.1145/3161601","abstract":{"ja":"We present in this paper a detailed comparison of different algorithms and devices to determine the number of words\nread in everyday life. We call our system the “Wordometer”. We used three kinds of eye tracking systems in our experiment:\nmobile video-oculography (MVoG); stationary video-oculography (SVoG); and electro-oculography (EoG). By analyzing the\nmovement of the eyes we were able to estimate the number of words that a user read. Recently, inexpensive eye trackers have\nappeared on the market. ?us, we undertook a large-scale experiment that compared three devices that can be used for daily\nreading on a screen: the Tobii Eye X SVoG; the JINS MEME EoG; and the Pupil MVoG. We found that the accuracy of the\neveryday life devices and professional devices was similar when used with the Wordometer. We analyzed the robustness of\nthe systems for special reading behaviors: rereading and skipping.\nWith the MVoG, SVoG and EoG systems, we obtained estimation errors respectively, 7.2%, 13.0%, and 10.6% in our main\nexperiment. In all our experiments, we obtained 300 recordings by 14 participants, which amounted to 109,097 read words."}},{"key":"Utsumi2017_IPSJ","id":"1567","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Yuzuko Utsumi","en":""},{"ja":"Tomoya Mizuno","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Fast Search Based on Generalized Similarity Measures"},"journal":{"ja":"IPSJ Transactions on Computer Vision and Applications"},"volume":"9","number":"1","num_pages":"6","year":2017,"month":3,"review":"reviewed","presen_type":"oral_and_poster","fields":["pattern-recognition"],"doi":"10.1186/s41074-017-0024-5","url":"http://link.springer.com/article/10.1186/s41074-017-0024-5","abstract":{"ja":"This paper proposes a fast recognition method based on generalized similarity measure (GSM). The GSM achieves good recognition accuracy for face recognition, but has a scalability problem. Because the GSM method requires the similarity measures between a query and all samples to be calculated, the computational cost for recognition is in proportion to the number of samples. A reasonable approach to avoiding calculating all the similarity measures is to limit the number of samples used for calculation. Although approximate nearest neighbor search (ANNS) methods take this approach, they cannot be applied to the GSM-based method directly because they assume that similarity measure is the Euclidean distance. The proposed method embeds the GSM into the Euclidean distance so that it may be applied in existing ANNS methods. We conducted experiments on face, object, and character datasets, and the results show that the proposed method achieved fast recognition without dropping the accuracy."}},{"key":"Iwamura2017_TRANS","id":"1536","type":"journal_paper","language":"EN","show":"show_except_datafile","authors":[{"ja":"","en":"Keiichi IWAMURA"},{"ja":"","en":"Masaki KAWAMURA"},{"ja":"","en":"Minoru KURIBAYASHI"},{"ja":"","en":"Motoi IWATA"},{"ja":"","en":"Hyunho KANG"},{"ja":"","en":"Seiichi GOHSHI"},{"ja":"","en":"Akira NISHIMURA"}],"title":{"en":"Information Hiding and Its Criteria for Evaluation"},"journal":{"en":"IEICE TRANS. INF. & SYST."},"volume":"100","number":"1","pages":"2-12","year":2017,"month":1,"fields":[],"url":"https://search.ieice.org/bin/summary.php?id=e100-d_1_2&category=-&lang=E&year=2017&abst=","abstract":{"en":"Within information hiding technology, digital watermarking is one of the most important technologies for copyright protection of digital content. Many digital watermarking schemes have been proposed in academia. However, these schemes are not used, because they are not practical; one reason for this is that the evaluation criteria are loosely defined. To make the evaluation more concrete and improve the practicality of digital watermarking, watermarking schemes must use common evaluation criteria. To realize such criteria, we organized the Information Hiding and its Criteria for Evaluation (IHC) Committee to create useful, globally accepted evaluation criteria for information hiding technology. The IHC Committee improves their evaluation criteria every year, and holds a competition for digital watermarking based on state-of-the-art evaluation criteria. In this paper, we describe the activities of the IHC Committee and its evaluation criteria for digital watermarking of still images, videos, and audio."}},{"key":"Iwata2017","id":"1537","type":"journal_paper","language":"EN","show":"show_except_datafile","authors":[{"ja":"","en":"Motoi IWATA"},{"ja":"","en":"Naoyoshi MIZUSHIMA"},{"ja":"","en":"Koichi KISE"}],"title":{"en":"Practical Watermarking Method Estimating Watermarked Region from Recaptured Videos on Smartphone"},"journal":{"en":"IEICE TRANS. INF. & SYST."},"volume":"100","number":"1","num_pages":"24-32","year":2017,"month":1,"review":"reviewed","fields":[],"doi":"10.1587/transinf.2016MUP0008","url":"https://search.ieice.org/bin/summary.php?id=e100-d_1_24","abstract":{"en":"In these days, we can see digital signages in many places, for example, inside stations or trains with the distribution of attractive promotional video clips. Users can easily get additional information related to such video clips via mobile devices such as smartphone by using some websites for retrieval. However, such retrieval is time-consuming and sometimes leads users to incorrect information. Therefore, it is desirable that the additional information can be directly obtained from the video clips. We implement a suitable digital watermarking method on smartphone to extract watermarks from video clips on signages in real-time. The experimental results show that the proposed method correctly extracts watermarks in a second on smartphone."}},{"key":"Augereau2017","id":"1594","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Olivier Augereau","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"An overview of comics researchin computer science"},"book_title":"Proceedings of 14th IAPR International Conference on Document Analysis and Recognition (ICDAR)","volume":"03","pages":"54-59","num_pages":"6","year":2017,"month":11,"review":"reviewed","fields":["document-image-analysis"],"abstract":{"ja":"A wide variety of research about comics has been proposed in the recent years. One of the main focuses of researchers in document image analysis is about comics content analysis and segmentation such as balloons, text or panels.\nResearch about user interaction or content generation has also been proposed. However, as these parts have been covered by different research fields (document image analysis, multimedia,\nartificial intelligence, human computer interaction, etc.) with\ndifferent sets of values. We propose in this paper an overview\nabout the previous research about comics in computer science.\nWe will state what have been done and which are the main\noutlooks of current research."}},{"key":"Daiku2017","id":"1595","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Yuki Daiku","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Comic story analysis based on genre classification"},"book_title":"Proceedings of 14th IAPR International Conference on Document Analysis and Recognition (ICDAR)","volume":"03","pages":"60-65","num_pages":"6","year":2017,"month":11,"publisher":"IEE","review":"reviewed","fields":["document-image-analysis"],"abstract":{"ja":"Comic readers are attracted to not only pictures of unique characters or beautiful landscape but also deliberated\nstory. Understanding comic story is helpful for a comic retrieval,\nwhich allows readers to obtain comics suited to readers’ interest,\nor creative activities, which demand to generate interesting idea\nof comic narratives. Therefore, as the method of understanding\ncomic story, we propose a converting method from a real comic\nstory into a novel formatted narrative structure, which uses comic\ngenres as the representation of contents of story. In a converting\nmethod, each page in a comic volume is classified into what\ngenre the page portrays by using convolutional neural network.\nGenerally, in machine learning, labeling ground truth on a large\nnumber of training samples is necessary, which spends much\ncosts of time and money. In this paper, we propose a learning\nway, which supports to label ground truth on such many samples.\nThe experimental results show the effectiveness of our proposed\nconverting method."}},{"key":"Sanches2017","id":"1576","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Charles Lima Sanches","en":"Charles Lima Sanches"},{"ja":"Olivier Augereau","en":"Olivier Augereau"},{"ja":"Koichi Kise","en":"Koichi Kise"}],"title":{"ja":"Using the eye gaze to predict document reading subjective understanding","en":"Using the eye gaze to predict document reading subjective understanding"},"journal":{"ja":"Proceedings of International Workshop on Human-Document Interaction"},"volume":"8","pages":"38-41","num_pages":"4","year":2017,"month":11,"publisher":"IEEE","review":"reviewed","fields":["reading-life-log"],"doi":"10.1109/ICDAR.2017.377","abstract":{"ja":"The traditional way to analyze the content of a document is to perform document image analysis.\nHowever, analyzing how the user perceives a document is another way to get information about the documents but also about the users.\nBy using sensors such as eye tracker, it is possible analyze the reader's skill or the document comprehensibility. \nIn this paper, we focus on predicting the user's understanding as it can be used as a feedback either for the author (to improve his document) or the user (to review the parts he did not understand). \n\nThe eye movements of the readers are recorded by an eye tracker while they read several documents, then several features are extracted and a support vector regression system is used to predict the readers' understanding.\nIn our experiment, 17 subjects were asked to read 19 documents for a total of 323 recordings.\nAs a first result, we prove that the subjective understanding of the reader can be predicted more accurately by using the eye gaze than by asking a multiple choice question.","en":"The traditional way to analyze the content of a document is to perform document image analysis.\nHowever, analyzing how the user perceives a document is another way to get information about the documents but also about the users.\nBy using sensors such as eye tracker, it is possible analyze the reader's skill or the document comprehensibility. \nIn this paper, we focus on predicting the user's understanding as it can be used as a feedback either for the author (to improve his document) or the user (to review the parts he did not understand). \n\nThe eye movements of the readers are recorded by an eye tracker while they read several documents, then several features are extracted and a support vector regression system is used to predict the readers' understanding.\nIn our experiment, 17 subjects were asked to read 19 documents for a total of 323 recordings.\nAs a first result, we prove that the subjective understanding of the reader can be predicted more accurately by using the eye gaze than by asking a multiple choice question."}},{"key":"Garain2017","id":"1593","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Utpal Garain","en":""},{"ja":"Onkar Pandit","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"Ayano Okoso","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Identification of Reader Specific Difficult Words by Analyzing Eye Gaze and Document Content"},"journal":{"ja":"Proceedings of the International Conference on Document Analysis and Recognition"},"num_pages":"6","year":2017,"month":11,"publisher":"IEEE","review":"reviewed","fields":["reading-life-log"],"abstract":{"ja":"This paper presents an approach for identifying reader specific difficult words while someone is reading a textual document. The work is motivated by the need of developing human-document interaction systems, in general and creating person-specific online educational content, in particular. Eye gaze information gives person specific behavior whereas textual content is analyzed to get general linguistic aspect of the document content. These two pieces of information are fused together through machine learning algorithms to identify the set of difficult words for a particular reader reading a particular document. An annotated dataset has been created where each word in a document is marked with its bounding box information and each reader identifies a set of difficult words while reading the document. The dataset consists of sixteen documents and each document is read by five subjects. The method is evaluated through recall-precision analysis. The impressive precision at high recall attests the feasibility of building a practical application based on this research. The experiment further brings out several interesting facts about human reading behavior."}},{"key":"Iwamura2017_ICDAR","id":"1599","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Naoyuki Morimoto","en":""},{"ja":"Keishi Tainaka","en":""},{"ja":"Dena Bazazian","en":""},{"ja":"Lluis Gomez","en":""},{"ja":"Dimosthenis Karatzas","en":""}],"title":{"ja":"ICDAR2017 Robust Reading Challenge on Omnidirectional Video"},"journal":{"ja":"Proc. 14th IAPR International Conference on Document Analysis and Recognition (ICDAR2017)"},"pages":"1448-1453","year":2017,"month":9,"location":"Kyoto, Japan","review":"reviewed","fields":["pattern-recognition"],"doi":"10.1109/ICDAR.2017.236","url":"http://rrc.cvc.uab.es/?ch=7","abstract":{"ja":"Results of ICDAR 2017 Robust Reading Challenge\non Omnidirectional Video are presented. This competition uses\nDowntown Osaka Scene Text (DOST) Dataset that was captured\nin Osaka, Japan with an omnidirectional camera. Hence, it\nconsists of sequential images (videos) of different view angles.\nRegarding the sequential images as videos (video mode), two\ntasks of localisation and end-to-end recognition are prepared.\nRegarding them as a set of still images (still image mode), three\ntasks of localisation, cropped word recognition and end-to-end\nrecognition are prepared. As the dataset has been captured in\nJapan, the dataset contains Japanese text but also include text\nconsisting of alphanumeric characters (Latin text). Hence, a\nsubmitted result for each task is evaluated in three ways: using\nJapanese only ground truth (GT), using Latin only GT and using\ncombined GTs of both. Finally, by the submission deadline, we\nhave received two submissions in the text localisation task of\nthe still image mode. We intend to continue the competition in\nthe open mode. Expecting further submissions, in this report\nwe provide baseline results in all the tasks in addition to the\nsubmissions from the community."}},{"key":"Ishimaru2017","id":"1697","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Shoya Ishimaru","en":""},{"ja":"Kensuke Hoshika","en":""},{"ja":"Kai Kunze","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Andreas Dengel","en":""}],"title":{"ja":"Towards reading trackers in the wild: detecting reading activities by EOG glasses and deep neural networks"},"journal":{"ja":"Proceedings of the 2017 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of the 2017 ACM International Symposium on Wearable Computers"},"pages":"704-711","year":2017,"month":9,"fields":["reading-life-log"],"doi":"10.1145/3123024.3129271","abstract":{"ja":"Reading in real life occurs in a variety of settings. One may read while commuting to work, waiting in a queue or lying on the sofa relaxing. However, most of current activity recognition work focuses on reading in fully controlled experiments. This paper proposes reading detection algorithms that consider such natural readings. The key idea is to record a large amount of data including natural reading habits in real life (more than 980 hours from 7 participants) with commercial electrooculography (EOG) glasses and to use them for deep learning. Our proposed approaches classified controlled reading vs. not reading with 92.2% accuracy on a user-dependent training. However, the classification accuracy decreases to 73.8% on natural reading vs. not reading. The results indicate that there is a strong gap between controlled reading and natural reading, highlighting the need for more robust reading detection algorithms."}},{"key":"Syeda2017","id":"1700","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Uzma Haque Syeda","en":""},{"ja":"Ziaul Zafar","en":""},{"ja":"Zishan Zahidul Islam","en":""},{"ja":"Syed Mahir Tazwar","en":""},{"ja":"Miftahul Jannat Rasna","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Md. Atiqur Rahman Ahad","en":""}],"title":{"ja":"Visual face scanning and emotion perception analysis between autistic and typically developing children"},"journal":{"ja":"Proceedings of the 2017 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of the 2017 ACM International Symposium on Wearable Computers"},"pages":"844-853","year":2017,"month":9,"review":"reviewed","fields":["sensing"],"doi":"10.1145/3123024.3125618","abstract":{"ja":"Autism is a neuro-developmental condition that is characterized by a number of unconventional behaviors such as restricted and repetitive activities. It is often largely attributed to deficiency in communication and social interaction. Therefore, it is difficult to make autistic individuals, especially children, to comply with researches that aim at comprehending this condition. However, with the availability of non-invasive eye-tracking technology, this problem has become easier to deal with. The following research probes into the visual face scanning patterns and emotion recognition between 21 autistic and 21 control or TD (typically developing) children when displayed pictures of 6 basic emotions (happy, sad, angry, disgusted, fearful and surprised). Tobii EyeX Controller was used to attain the gaze data and the data was processed and analyzed in MATLAB. The results revealed that children with autism look less at the core features of the face (eyes, nose and mouth) while scanning faces and have more difficulty in perceiving the correct emotion compared to the typically developing children. This atypical face scanning and lack of preference to the core features of the face can be the reason why autistic individuals have trouble understanding others' emotions and an overall incompetency in communication and social interaction. To delve more into this, further eye-tracking, neuroimaging and behavioral studies should be done in integration."}},{"key":"Hitinui2017","id":"1589","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Robert Hitinui","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Real-time wordometer demonstration using commercial EoG glasses"},"journal":{"ja":"Proceedings of the 2017 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of the 2017 ACM International Symposium on Wearable Computers (UbiComp'17)"},"pages":"277-280","num_pages":"4","year":2017,"month":9,"publisher":"ACM","review":"not_reviewed","fields":["reading-life-log"],"doi":"10.1145/3123024.3123183","abstract":{"ja":"Reading is an important part of our everyday life. Most of us read every day at work, in the transportation, at home, etc. Except by counting the number of books a person read in a year or a month, it is very hard to quantify reading. We want to create a \"wordometer\" that counts how many words a user read. Such an application could encourage people to read more, just like pedometers encourage to walk more. We propose a demonstration of such a system based on eye movements analysis extracted from EoG glasses."}},{"key":"Kise2017_UbiComp","id":"1610","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"","en":"Koichi Kise"},{"ja":"","en":"Olivier Augereau"},{"ja":"","en":"Yuzuko Utsumi"},{"ja":"","en":"Masakazu Iwamura"},{"ja":"","en":"Kai Kunze"},{"ja":"","en":"Shoya Ishimaru"},{"ja":"","en":"Andreas Dengel"}],"title":{"en":"Quantified reading and learning for sharing experiences"},"journal":{"en":"Proceedings of the 2017 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of the 2017 ACM International Symposium on Wearable Computers (UbiComp'17)"},"pages":"724-731","num_pages":"8","year":2017,"month":9,"publisher":"ACM","location":"Maui, Hawaii","review":"reviewed","fields":[],"doi":"10.1145/3123024.3129274","url":"https://dl.acm.org/citation.cfm?id=3129274"},{"key":"Yamada2017_UbiComp","id":"1590","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Kento Yamada","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Estimation of confidence based on eye gaze: an application to multiple-choice questions"},"journal":{"ja":"Proceedings of the 2017 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of the 2017 ACM International Symposium on Wearable Computers (UbiComp'17)"},"pages":"217-220","num_pages":"4","year":2017,"month":9,"publisher":"ACM","review":"not_reviewed","fields":["reading-life-log"],"doi":"10.1145/3123024.3123138","abstract":{"ja":"Multiple-choice questions are often used to assess the user's understanding. However, the system does not know if the user answered randomly, by chance, with confidence or without confidence. The user's confidence can be used as a feedback for teachers or smart interfaces that could understand that he is not sure of himself and then provide further assistance.\n\nWe aim to create confidence aware interfaces such as pervasive display. As a first step, we propose an application in the context of multiple-choice question answering. By using an eye tracker, we analyze the reading and answering behavior of the user, extract some features and predict his confidence. We asked 11 participants to solve 80 multiple-choice questions about English understanding, and estimated whether they answered confidently or not. We obtained 90.1% average accuracy for this prediction."}},{"key":"Sanches2017_ACM","id":"1575","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Charles Lima Sanches","en":"Charles Lima Sanches"},{"ja":"Olivier Augereau","en":"Olivier Augereau"},{"ja":"Koichi Kise","en":"Koichi Kise"}],"title":{"ja":"Japanese Reading Objective Understanding Estimation by Eye Gaze Analysis","en":"Japanese Reading Objective Understanding Estimation by Eye Gaze Analysis"},"journal":{"ja":"Proceedings of the 2017 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of  the 2017 ACM International Symposium on Wearable Computers (UbiComp'17)","en":"Proceedings of the 2017 ACM International Joint Conference on"},"series":"UbiComp '17","pages":"121-124","num_pages":"4","year":2017,"month":8,"publisher":"ACM","address":"New York, NY, USA","location":"Maui, Hawaii","review":"reviewed","fields":["reading-life-log","learning-analytics"],"doi":"10.1145/3123024.3123092","abstract":{"ja":"Reading is an important part of our everyday life. Most of us read every day at work, in the transportation, at home, etc. Except by counting the number of books a person read in a year or a month, it is very hard to quantify reading. We want to create a \"wordometer\" that counts how many words a user read. Such an application could encourage people to read more, just like pedometers encourage to walk more. We propose a demonstration of such a system based on eye movements analysis extracted from EoG glasses.","en":"Analyzing the eye gaze to estimate the text understanding is a good way to overcome the drawbacks of classic question based assessment tests.\nIn particular it does not suffer from random answers, question misunderstanding and can cover every parts of the text.\nIn this paper we propose a method to estimate the objective understanding of a learner by analyzing his eye movements while reading.\nWe conduct our experiment on Japanese texts and try to predict, by analyzing the eye gaze, how many questions about the texts the reader will be able to answer to.\nWe show that we obtain 5.27% of error in the number of correct answers estimation by using eye gaze features. \nAs a comparison, we try to predict the number of correct answers by using the reader's self assessment understanding and show that it leads to 9.04% error."}},{"key":"Kise2017_PLATO","id":"1574","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":"Koichi Kise"}],"title":{"ja":"Deeply Sensing Learners for Better Assistance: Towards Distribution of Learning Experiences","en":"Deeply Sensing Learners for Better Assistance: Towards Distribution of Learning Experiences"},"journal":{"ja":"Proceedings of the PLATO (Positive Learning in the age of information) conference","en":"Proceedings of the PLATO (Positive Learning in the age of information) conference"},"year":2017,"month":6,"review":"not_reviewed","fields":["reading-life-log"],"abstract":{"ja":"Most of the current e-learning systems rely on shallow sensing of learners such as achievement tests and log of usage of e-learning systems. This poses a limitation to know internal states of learners such as confidence and the level of knowledge. To solve this problem, we propose to employ deeper sensing by using eye trackers, EOG, EEG, motion and physiological sensors. As tasks, we consider English learning. The sensing technologies described in this abstract includes low level estimations (the number of read words, the period of reading), document type recognition and identification of read words, as well as high level estimations about confidence of answers, the English ability in terms of TOEIC scores and unknown words encountered while reading English documents. Such functionality helps learners and teachers to know the internal states and will be used to describe learning experiences to be shared by other learners.","en":"Most of the current e-learning systems rely on shallow sensing of learners such as achievement tests and log of usage of e-learning systems. This poses a limitation to know internal states of learners such as confidence and the level of knowledge. To solve this problem, we propose to employ deeper sensing by using eye trackers, EOG, EEG, motion and physiological sensors. As tasks, we consider English learning. The sensing technologies described in this abstract includes low level estimations (the number of read words, the period of reading), document type recognition and identification of read words, as well as high level estimations about confidence of answers, the English ability in terms of TOEIC scores and unknown words encountered while reading English documents. Such functionality helps learners and teachers to know the internal states and will be used to describe learning experiences to be shared by other learners."}},{"key":"Utsumi2017_MVA","id":"1578","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Yuzuko Utsumi","en":""},{"ja":"Tomoya Mizuno","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Fast Search Based on Generalized Similarity Measures"},"journal":{"ja":"Proc. Fifteenth IAPR International Conference on Machine Vision Applications (MVA)"},"year":2017,"month":5,"location":"Nagoya, Japan","review":"not_reviewed","fields":["pattern-recognition"],"doi":"10.23919/MVA.2017.7986831","awards":{"en":["MVA 2017 Best Paper Award Winner"]},"abstract":{"ja":"This paper proposes a fast recognition method based on generalized similarity measure (GSM). The GSM achieves good recognition accuracy for face recognition, but has a scalability problem. Because the GSM method requires the similarity measures between a query and all samples to be calculated, the computational cost for recognition is in proportion to the number of samples. A reasonable approach to avoiding calculating all the similarity measures is to limit the number of samples used for calculation. Although approximate nearest neighbor search (ANNS) methods take this approach, they cannot be applied to the GSM-based method directly because they assume that similarity measure is the Euclidean distance. The proposed method embeds the GSM into the Euclidean distance so that it may be applied in existing ANNS methods. We conducted experiments on face, object, and character datasets, and the results show that the proposed method achieved fast recognition without dropping the accuracy."}},{"key":"Nakamura2017_SBRA","id":"1596","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"中村浩一朗","en":"Koichiro NAKAMURA"},{"ja":"内海ゆづ子","en":"Yuzuko UTSUMI"},{"ja":"岩村雅一","en":"Masakazu IWAMURA"},{"ja":"黄瀬浩一","en":"Koichi KISE"},{"ja":"槇原靖","en":"Yasushi MAKIHARA"},{"ja":"村松大吾","en":"Daigo MURAMATSU"},{"ja":"八木康史","en":"Yasushi YAGI"}],"title":{"ja":"歩容認証の高精度化のための人物領域の抽出手法の改良","en":"Improvement on Individuality-Preserving Silhouette Extraction for High Accurate Gait Recognition"},"journal":{"ja":"第7回バイオメトリクスと認識・認証シンポジウム(SBRA2017)"},"presen_id":"SBRA2017-S2-17","pages":"43-44","num_pages":"2","year":2017,"month":11,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"防犯カメラによる犯罪捜査への応用を\n目的とした歩容認証が注目を集めている．歩容認証の手\n法は，画像上の人物領域を抽出して得られる歩容シルエッ\nト画像を用いる方法が現在の主流となっている．人物領\n域を手動で抽出すると負担がかかるため，自動で抽出す\nる手法として，標準歩容モデルを用いた人物領域抽出が\n提案されている．この手法では，抽出する対象となる人\n物と類似する標準歩容モデルを選択し，適用することで，\n歩容の認証精度が向上することが示されている．しかし，\n人物領域の抽出精度の評価は限定的である．そこで，本\n研究では，標準歩容モデルの適用により，人物領域の抽\n出精度がどのように変化するかを調査した．また，類似\nする標準歩容モデルの選択方法の改善により，人物領域\nの抽出精度を更に向上させる手法を提案する．実験にお\nいて，提案手法を用いたときが最も高い人物領域抽出精\n度を示した．"}},{"key":"Yamada2017_PRMUc","id":"1581","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"山田 良博","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"PyramidNetに対する新たな確率的正則化手法ShakeDropの提案"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"117","number":"238","presen_id":"PRMU2017-72","pages":"55-60","num_pages":"6","year":2017,"month":10,"location":"熊本大学, 熊本県熊本市","review":"not_reviewed","fields":["object-recognition","deep-learning"],"awards":{"ja":["2017年度 PRMU研究奨励賞受賞"],"en":["2017 PRMU Award Winner"]},"abstract":{"ja":"高精度な一般物体認識を実現するために， 確率的に学習を乱すことで認識精度を向上する，確率的な正則化手法が注目を集めている． 確率的な正則化を用いた一般物体認識手法の1つであるPyramidDropは，Randam Dropと呼ばれる確率的な正則化を導入することで発表当時の世界最高性能を実現した． しかし，Randam Dropは二値の乱数を用いて各Residual Unitを使うか使わないかを制御するため，正則化の効果が限定的と考えられる． 本稿では，最近提案された実数の乱数を用いる確率的な正則化をPyramidDropに導入することで，より高い認識精度の手法の実現を目指す． この乱数は，各Residual Unitをどの程度使うかを表す． 一般物体認識データセットCIFAR-10及びCIFAR-100を用いた実験の結果，提案手法ShakeDropは，現在のstate-of-the-artの手法に比べて、エラー率をCIFAR-10で0.25\\%、CIFAR-100で3.01\\%軽減して，現時点の世界最高精度を達成することを確認した．"}},{"key":"Yamada2017_MIRUa","id":"1577","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"山田 健斗","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"視線に基づく英語多肢選択問題解答の確信度推定"},"journal":{"ja":"画像の認識理解シンポジウム(MIRU2017)"},"num_pages":"4","year":2017,"month":8,"location":"広島国際会議場","review":"not_reviewed","fields":["reading-life-log"]},{"key":"Yamada2017_MIRUb","id":"1579","type":"domestic_conf","language":"EN","show":"show","authors":[{"ja":"Yoshihiro Yamada","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Deep Pyramidal Residual Networks with Stochastic Depth"},"journal":{"ja":"第20回画像の認識・理解シンポジウム(MIRU2017)予稿集"},"year":2017,"month":8,"location":"広島県広島市","review":"not_reviewed","fields":[],"awards":{"ja":["MIRUインタラクティブ発表賞受賞"]}},{"key":"Kise2017_WS","id":"1573","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"Olivier Augereau","en":"Olivier Augereau"},{"ja":"Charles Lima Sanches","en":"Charles Lima Sanches"},{"ja":"藤好 宏樹","en":"Hiroki Fujiyoshi"},{"ja":"大社 綾乃","en":"Ayano Okoso"},{"ja":"山田 健斗","en":"Kento Yamada"},{"ja":"Kai Kunze","en":"Kai Kunze"},{"ja":"石丸 翔也","en":"Shoya Ishimaru"},{"ja":"Andreas Dengel","en":"Andreas Dengel"}],"title":{"ja":"様々なセンサを用いた読書行動解析","en":"Analysis of Reading Activities by Using Various Sensors"},"journal":{"ja":"教育情報システム学会研究会報告","en":"JSiSE Research Report"},"volume":"32","number":"2","pages":"53-60","year":2017,"month":7,"location":"長野県松本市","review":"not_reviewed","fields":["reading-life-log"],"abstract":{"ja":"Most of the current e-learning systems rely on shallow sensing of learners such as achievement tests and log of usage of the systems. This poses a limitation to know internal states of learners such as confidence and the level of knowledge. To solve this problem, we propose to employ deeper sensing by using eye trackers, EOG, EEG, motion and physiological sensors. As tasks, we consider learning of English and Japanese. The sensing technologies described in this report includes low level estimations (the number of read words, the period of reading), document type recognition and identification of read words, as well as high level estimations about confidence of answers, the English ability in terms of TOEIC scores, unknown words encountered while reading English documents, and subjective and objective level of understanding of Japanese. Such functionality helps learners and teachers to know the internal states.","en":"Most of the current e-learning systems rely on shallow sensing of learners such as achievement tests and log of usage of the systems. This poses a limitation to know internal states of learners such as confidence and the level of knowledge. To solve this problem, we propose to employ deeper sensing by using eye trackers, EOG, EEG, motion and physiological sensors. As tasks, we consider learning of English and Japanese. The sensing technologies described in this report includes low level estimations (the number of read words, the period of reading), document type recognition and identification of read words, as well as high level estimations about confidence of answers, the English ability in terms of TOEIC scores, unknown words encountered while reading English documents, and subjective and objective level of understanding of Japanese. Such functionality helps learners and teachers to know the internal states."}},{"key":"Shibayama2017_CVIM","id":"1569","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"柴山 祐輝","en":""},{"ja":"森本 直之","en":""},{"ja":"山田 良博","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"大規模なデータセットを構築するための画像のフィルタリング手法"},"journal":{"ja":"研究報告コンピュータビジョンとイメージメディア(CVIM)"},"volume":"2017-CVIM-207","number":"30","pages":"1-5","year":2017,"month":5,"location":"名古屋大学","review":"not_reviewed","fields":[],"url":"https://ipsj.ixsq.nii.ac.jp/ej/?action=pages_view_main&active_action=repository_view_main_item_detail&item_id=178763&item_no=1&page_id=13&block_id=8"},{"key":"Nakamura2017_CVIM","id":"1568","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"中村 浩一朗","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""},{"ja":"槇原 靖","en":""},{"ja":"村松 大吾","en":""},{"ja":"八木 康史","en":""}],"title":{"ja":"歩容認証の高精度化のための人物領域の抽出精度の評価"},"journal":{"ja":"研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2017-CVIM-207","number":"18","year":2017,"month":5,"location":"名古屋大学","review":"not_reviewed","fields":[]},{"key":"Miyamoto2017","id":"1572","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"宮本 康平","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"LLAH をノイズおよび距離変化に頑健にする手法の提案"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"117","number":"49","presen_id":"PRMU2017-23","pages":"119-124","num_pages":"6","year":2017,"month":5,"location":"名古屋工業大学","review":"not_reviewed","fields":["document-image-analysis"],"awards":{"ja":["2017年5月の月間ベストプレゼンテーション賞受賞"],"en":["May 2017 PRMU Best Presentation Award Winner"]},"abstract":{"ja":"スマートフォンの普及により，カメラは誰しもが持つデバイスとなった．同時に，そのカメラで撮影された画像上にリアルタイムに情報を付加するサービスが普及し，それを実現する技術の一つにカメラベースの文書画像検索がある．これは，カメラで撮影した文書画像と同じ文書を，予め登録しておいた文書画像の中から検索するというものである．Locally Likely Arrangement Hashing (LLAH) は，主に単語領域の重心を特徴点とし，そこから計算された幾何学的不変量を用いることで，高速な検索を実現した．しかし， LLAH の特徴点抽出法はノイズに弱いという欠点があった． LLAH では，同じ単語内の文字だけが連結するように文字をぼかし，単語領域を得ていた．ここで，適切なぼかし具合は，文字の大きさによって変化するため，文字の面積からぼかし具合の推定を行っていた．しかし，ノイズがその推定に影響を与えるという問題があった．本稿では，画像から文字の成分を取り出し，その面積から文字とノイズを EM アルゴリズムによって動的に分離し，ノイズに頑健なぼかし具合の推定を行った．また，一定のぼかし具合であっても，画像を拡大縮小することによって，ぼかし具合を変化させたときと，同じ効果を得ることができる．これにより，推定されたぼかし具合から，画像の大きさを正規化することで，特徴点抽出の精度と速度が向上させた．"}},{"key":"Yamamoto2017_PRMU","id":"1547","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"山本大貴","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"次元数の異なる部分空間同士の近似最近傍部分空間探索"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"116","number":"528","presen_id":"PRMU2016-199","pages":"17-22","year":2017,"month":3,"location":"名城大学","review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"パターンを部分空間で表現する手法がパターン認識やコンピュータビジョンの分野において広く利用されている．\n部分空間を使ってクエリのパターンを認識する際，部分空間同士の類似度を計算する．この部分空間同士の類似度計算は計算コストが高く，1対1の計算をする必要があるため，データベースが大きいと処理時間が膨大となってしまう．\n扱う問題の大規模化に伴い，類似度計算の近似，そして探索対象の効率的な絞り込みが必要とされている．そのような近似手法にグラスマン多様体が用いられる．グラスマン多様体は，同じ次元数の部分空間の集合であり，グラスマン多様体上において，部分空間は点で表現される．このことから，グラスマン多様体上での近似最近傍探索によって，探索候補を絞り込み，高速な部分空間の探索手法が提案されている．しかし，グラスマン多様体は，次元数の異なる部分空間は扱えず，このような高速化手法が適用できる問題は限られる．また，次元数の異なる部分空間を扱う従来の手法では，特異値分解のような高コストな計算が必要であり，高速化を阻む要因となると考えられる．本稿では，次元数の異なる部分空間同士のグラスマン距離について検討を行い，次元数の異なる部分空間を同じグラスマン多様体上での類似度比較を近似的に実現し，従来の高速化手法での探索を可能にする．"}},{"key":"Yamada2017_PRMUb","id":"1542","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"山田 良博","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"PyramidNetにおける確率的な正則化の効果の検証"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"116","number":"461","presen_id":"PRMU2016-158","pages":"35-40","year":2017,"month":2,"location":"北海道大学，北海道札幌市","review":"not_reviewed","fields":["object-recognition"],"awards":{"ja":["2016年度PRMU研究奨励賞受賞","2017年2月の月間ベストプレゼンテーション賞受賞"],"en":["2016 PRMU Award Winner","February 2017 PRMU Best Presentation Award Winner"]},"abstract":{"ja":"画像を『山』『椅子』『ラーメン』といったカテゴリで分類する問題を一般物体認識と呼ぶ．しかし高精度な一般物体認識は膨大な枚数の画像での学習を行うことが一般的であり，データセットの作成コストや学習時間の観点から現実的でない．そこで限られた枚数でも高い精度を実現する一般物体認識手法が求められている．限られたデータ量でも高い精度を実現する一般物体認識手法の1つがPyramidNetである．PyramidNetは従来，幾つかの段階で行われていた処理を逐次的に行うことによって，従来手法から飛躍的に認識精度を高めた手法である．しかしPyramidNetは従来手法で精度を高めるために用いられていた確率的な正則化処理を含んでおらず，性能が頭打ちになっている可能性がある．本稿ではPyramidNet に適した確率的な正則化手法を検討するべく評価実験を行った．"}},{"key":"Fujiyoshi2017_PRMU","id":"1541","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"藤好 宏樹","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"問題解答時の視点情報を用いた文書非依存な英語能力推定法"},"journal":{"ja":"電子情報通信学会技術研究報告, パターン認識メディア理解研究会"},"volume":"116","number":"461","presen_id":"PRMU2016-153","pages":"7-12","num_pages":"6","year":2017,"month":2,"location":"北海道大学","review":"not_reviewed","fields":[]},{"key":"Okoso2017_PRMU","id":"1538","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"大社 綾乃","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"視点情報と単語の出現頻度を用いた主観的高難易度単語の推定"},"journal":{"ja":"電子情報通信学会技術研究報告, パターン認識メディア理解研究会"},"presen_id":"PRMU2016-190","pages":"187-192","num_pages":"6","year":2017,"month":2,"location":"北海道大学","review":"not_reviewed","fields":["reading-life-log"],"abstract":{"ja":"本稿では，視点情報と単語の出現頻度情報を組み合わせた，文書上の主観的高難易度単語の推定手法を提 案する．「主観的高難易度単語」とは，文書を読んだ際にユーザが難しいと感じた単語を表す．文書を読むだけで， 主観的高難易度単語を推定できれば，ユーザはどの単語を学習しなければならないかを簡単に把握することが可能 になり，学習補助に大変有用である．一般的に，難しい単語を目にすると，読む速度が遅くなる，前後の文を読み 返すなどの目の動きがみられる．提案手法では，視点情報と単語情報から主観的高難易度単語の推定を目指す．ま ずアイトラッカから取得した視点情報より特徴量を抽出し，単語情報と組み合わせて入力データを作成する．Deep Learningを用いて学習し，主観的高難易度単語を推定する．提案手法の有効性を示すため，５名の視点情報を記録 し，主観的高難易度単語の推定精度を検証した．検証の結果，全単語を対象とした場合，再現率が最大のとき，再 現率99.0%，適合率は33.4%であった．また，固有名詞を学習及び評価時に除いた場合は，再現率が最大のとき，再 現率99.0%，適合率は39.6%であった．これは比較手法である，出現頻度情報のみを用いた手法よりも勝る結果と なり，提案手法の有効性を示すことができた．"}},{"key":"Yamada2017_PRMUa","id":"1540","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"山田 健斗","en":""},{"ja":"大社 綾乃","en":""},{"ja":"藤好 宏樹","en":""},{"ja":"星加 健介","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"英語多肢選択問題解答時の視線に基づく確信度推定"},"journal":{"ja":"電子情報通信学会技術研究報告, パターン認識メディア理解研究会"},"presen_id":"PRMU2016-192","pages":"199-204","num_pages":"6","year":2017,"month":2,"location":"北海道大学","review":"not_reviewed","fields":["reading-life-log"],"abstract":{"ja":"英語学習において，復習は重要な学習プロセスである．\n 十分な復習を行うためには，間違えた問題だけでなく，解答が正解であると確信をもてなかった問題も見落としてはならない．\nそこで本研究では，問題解答時の確信度を推定し，解答の正誤と合わせて\n解答者に提示するシステムの実現を目指す．このシステムの実現によって，見落としがなく，効率的で優先順位を考慮した復習が可能となる．\n 人は読んでいる文書を難しいと感じると，読む速度が遅くなる，読み返しが起こるといった特徴が見られる．このため，確信度の推定には視点情報を利用する．\n 本稿では，被験者11名に文法や語彙を問う英語の四択問題に解答してもらい，その際の問題に対する確信の有無について推定を行った．\n 実験においては，被験者依存で学習を行い，推定にSVMを使用した．\nその結果，被験者11名の平均推定精度は89.8%であった．"}},{"key":"Kise2017a","id":"1570","type":"book","language":"EN","show":"show_except_datafile","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Masahiko Inami","en":""}],"title":{"ja":"Reading-Life Log as a New Paradigm of Utilizing Character and Document Media"},"book_title":"Human-Harmonized Information Technology","volume":"2","year":2017,"month":4,"publisher":"Springer","review":"not_reviewed","fields":["reading-life-log","document-image-analysis"],"doi":"10.1007/978-4-431-56535-2_7","abstract":{"ja":"“You are what you read.” As this sentence implies, reading is important for building our minds. We are investing a huge amount of time for reading to input information. However the activity of “reading” is done only by each individual in an analog way and nothing is digitally recorded and reused. In order to solve this problem, we record reading activities as digital data and analyze them for various goals. We call this research “reading-life log.” In this chapter, we describe our achievements of the reading-life log. A target of the reading-life log is to analyze reading activities quantitatively and qualitatively: when, how much, what you read, and how you read in terms of your interests and understanding. Body-worn sensors including intelligent eyewear are employed for this purpose. Another target is to analyze the contents of documents based on the users’ reading activities: for example, which are the parts most people feel difficult/interesting. Materials to be read are not limited to books and documents. Scene texts are also important materials which guide human activities."}},{"key":"Mizuno2017","id":"1780","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"水野 智也","en":""}],"title":{"ja":"Generalized Similarity Measure のユークリッド距離への埋め込みによる認識の高速化","en":"Similarity Measure Embedding Method into Euclidean Space for Fast Recognition"},"year":2017,"month":3,"fields":[]},{"key":"Matsushita2017","id":"1624","type":"other","language":"JA","show":"show_except_datafile","authors":[{"ja":"松下光範","en":""},{"ja":"山西良典","en":""},{"ja":"松井勇佑","en":""},{"ja":"岩田基","en":""},{"ja":"上野未貴","en":""},{"ja":"西原陽子","en":""},{"ja":"中村聡史","en":""}],"title":{"ja":"私のブックマーク「コミック工学（Comic Computing）」"},"journal":{"ja":"人工知能学会誌"},"volume":"32","number":"6","pages":"999-1007","year":2017,"month":11,"review":"not_reviewed","fields":[]},{"key":"Kise2017b","id":"1571","type":"other","language":"EN","show":"show_except_abstract","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Masahiko Inami","en":""},{"ja":"Kai Kunze","en":""}],"title":{"ja":"Reading-Life Log as a New Paradigm of Utilizing Character and Document Media"},"book_title":"Human-Harmonized Information Technology, Volume 2 (Editor: Toyoaki Nishida)","pages":"197-234","year":2017,"month":5,"publisher":"Springer","review":"not_reviewed","fields":["reading-life-log","document-image-analysis"],"doi":"10.1007/978-4-431-56535-2_7"},{"key":"Sanches2016_IPSJ","id":"1532","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Charles Lima Sanches","en":"Charles Lima Sanches"},{"ja":"Olivier Augereau","en":"Olivier Augereau"},{"ja":"Koichi Kise","en":"Koichi Kise"}],"title":{"ja":"Vertical error correction of eye trackers in nonrestrictive reading condition","en":"Vertical error correction of eye trackers in nonrestrictive reading condition"},"journal":{"ja":"IPSJ Transactions on Computer Vision and Applications","en":"IPSJ Transactions on Computer Vision and Applications"},"volume":"8","number":"1","num_pages":"7","year":2016,"month":9,"publisher":"Springer Berlin Heidelberg","review":"reviewed","fields":[],"doi":"10.1186/s41074-016-0008-x","abstract":{"ja":"The eye tracking technology is used for four decades for studying reading behavior. The applications are various: estimating the reader comprehension, identifying the reader, summarizing a read document, creating a reading-life log, etc. The gaze data used in such applications has to be accurate enough to perform the analysis. In order to improve the accuracy, most of the experiments are set up with restrictive conditions such as using a head fixation and a professional eye tracker. It implies that the results are valid only in restrictive laboratory settings and an unrealistic small error is produced by the experiment. However, the use of affordable eye trackers in realistic conditions of reading leads to large errors in the recordings. We propose a new algorithm to correct the vertical error and to align the gazes with the text. The proposed algorithm is robust to rereading and skipping some parts of text, contrary to all the other algorithms of the state of the art. We show that up to 69% of the gazes are aligned with the correct text lines.","en":"The eye tracking technology is used for four decades for studying reading behavior. The applications are various: estimating the reader comprehension, identifying the reader, summarizing a read document, creating a reading-life log, etc. The gaze data used in such applications has to be accurate enough to perform the analysis. In order to improve the accuracy, most of the experiments are set up with restrictive conditions such as using a head fixation and a professional eye tracker. It implies that the results are valid only in restrictive laboratory settings and an unrealistic small error is produced by the experiment. However, the use of affordable eye trackers in realistic conditions of reading leads to large errors in the recordings. We propose a new algorithm to correct the vertical error and to align the gazes with the text. The proposed algorithm is robust to rereading and skipping some parts of text, contrary to all the other algorithms of the state of the art. We show that up to 69% of the gazes are aligned with the correct text lines."}},{"key":"Dingler2016","id":"1803","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Tilman Dingler","en":"Tilman Dingler"},{"ja":"Ioannis Giannopolos","en":"Ioannis Giannopolos"},{"ja":"Kai Kunze","en":"Kai Kunze"},{"ja":"Andreas Dengel","en":"Andreas Dengel"},{"ja":"Evangelos Niforatos","en":"Evangelos Niforatos"},{"ja":"Koichi Kise","en":"Koichi Kise"},{"ja":"Cathal Gurrin","en":"Cathal Gurrin"}],"title":{"ja":"WAHM 2016: 3rd workshop on ubiquitous technologies for augmenting the human mind"},"journal":{"en":"UbiComp 2016 Adjunct - Proceedings of the 2016 ACM International Joint Conference on Pervasive and Ubiquitous Computing"},"pages":"1010","num_pages":"1013","year":2016,"month":9,"fields":[],"doi":"10.1145/2968219.2968570"},{"key":"Shiga2016_IEICE","id":"1521","type":"journal_paper","language":"JA","show":"show_except_datafile","authors":[{"ja":"志賀 優毅","en":"Yuki SHIGA"},{"ja":"内海 ゆづ子","en":"Yuzuko UTSUMI"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"},{"ja":"Kai Kunze","en":"Kai KUNZE"},{"ja":"黄瀬 浩一","en":"Koichi KISE"}],"title":{"ja":"視線情報と一人称視点画像を用いた文書カテゴリの自動識別","en":"Automatic Document Type Classification Using Eye Movements and EgocentricImages"},"journal":{"ja":"電子情報通信学会論文誌","en":"Trans. IEICE"},"volume":"J99-D","number":"9","pages":"950-958","year":2016,"month":9,"review":"reviewed","fields":["reading-life-log"],"doi":"10.14923/transinfj.2015JDP7079","url":"http://search.ieice.org/bin/summary.php?id=j99-d_9_950","abstract":{"ja":"近年，人間の生活をデジタルデータとして記録するライフログが注目されている．ライフログの中には，読書活\n動の記録を目的としたReading-Life Log と呼ばれるものがある．現在，読書活動の記録は主に手動で行われて\nいるため，記録に手間がかかる．読書活動の自動的な記録が一般的になれば，より手軽に読書活動を記録できる\nようになる．読書活動の自動的な記録を目的として，我々はウェアラブルカメラで撮影された一人称視点画像を\n用いて，文書のカテゴリを識別した．しかし，この手法には文書領域を切り出す際，手動の手順を含むという問\n題があった．そこで，本論文ではモバイルアイトラッカから得られる視線情報に基づいて文書領域を自動的に切\nり出す手法を提案する．さらに，画像特徴量に加え，視線特徴量を併用することで識別精度を改善する．実験の\n結果，提案手法は従来手法と同程度の識別精度を保ったまま，自動化を達成した．","en":"Reading-Life Log aims to record reading activities. Recording reading activities manually is\nlaborious. If reading activities can be recorded automatically, people will be able to record reading activities\nmore easily. Though we classified document categories by using egocentric images recorded with a\nwearable camera for the purpose of recording reading activities automatically, it requires manual cropping\nof document regions. In this paper, we propose an automatic document classification method. It uses not\nonly image features but gaze features to increase classification accuracy."}},{"key":"Utsumi2016","id":"1486","type":"letter","language":"JA","show":"show","authors":[{"ja":"内海 ゆづ子","en":"Yuzuko Utsumi"},{"ja":"勝手 美紗","en":"Misa Katte"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"物体検出器の検出傾向に基づく動画像のイベント検出"},"journal":{"ja":"電子情報通信学会論文誌D","en":"IEICE TRANSACTIONS on Information and Systems"},"volume":"J99-D","number":"5","pages":"583-587","year":2016,"month":5,"review":"reviewed","fields":["pattern-recognition"],"doi":"10.14923/transinfj.2015JDL8020","url":"http://search.ieice.org/bin/summary.php?id=j99-d_5_583"},{"key":"Augereau2016a","id":"1535","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Olivier Augereau","en":""},{"ja":"Mizuki Matsubara","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Comic visualization on smartphones based on eye tracking"},"book_title":"Proceedings of the 1st International Workshop on coMics ANalysis, Processing and Understanding","pages":"4:1--4:4","num_pages":"4","year":2016,"month":12,"publisher":"ACM","review":"reviewed","fields":[]},{"key":"Augereau2016b","id":"1543","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Olivier Augereau","en":""},{"ja":"Hiroki Fujiyoshi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Towards an Automated Estimation of English Skill via TOEIC Score Based on Reading Analysis"},"book_title":"Proceedings of the 23rd International Conference on Pattern Recognition (ICPR)","pages":"1285-1290","num_pages":"6","year":2016,"month":12,"review":"reviewed","fields":["learning-analytics","reading-life-log"],"abstract":{"ja":"Estimating automatically the degree of language skill by analyzing the eye movements is a promising way to help people from all over the world to learn a new language. In this study, we focus on the English skills of non-native speakers. Our aim is to provide an algorithm that can assess accurately and automatically the TOEIC score after reading English texts for few minutes. As a first step towards this direction, we propose an algorithm that can predict accurately this score after reading and answering some questions about the comprehension of few English texts. We use an eye tracker in order to record the eye gaze, i.e. the positions where the reader is looking at. Then\nwe extract several features to characterize the behavior, and consequently the skill of the reader. We also add a feature based on the number of correct answers to the questions. By using a machine learning based on multivariate regression, the score is estimated user independently. A backward stepwise feature selection is used to select the relevant features and to optimize the estimation. As a main result, the TOEIC score is estimated with 21.7 points of mean absolute error for 21 subjects after reading and answering the questions of only 3 documents."}},{"key":"Matsubara2016","id":"1531","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Mizuki Matsubara","en":"Mizuki Matsubara"},{"ja":"Olivier Augereau","en":"Olivier Augereau"},{"ja":"Charles Lima Sanches","en":"Charles Lima Sanches"},{"ja":"Koichi Kise","en":"Koichi Kise"}],"title":{"ja":"Emotional arousal estimation while reading comics based on physiological signal analysis","en":"Emotional arousal estimation while reading comics based on physiological signal analysis"},"journal":{"ja":"Proceedings of the 1st International Workshop on coMics ANalysis, Processing and Understanding"},"num_pages":"4","year":2016,"month":12,"location":"Cancun, Mexico","review":"reviewed","fields":["reading-life-log","learning-analytics"],"doi":"10.1145/3011549.3011556","abstract":{"ja":"Estimating emotions while reading enables new services such as comic recommendation. Most of existing emotion estima- tion systems employ bulky devices. Furthermore, few ap- plications have been developed for analyzing the emotions while reading. The purpose of our research is to develop a method for estimating emotions while reading. As the tar- get of reading, we select comics, which stimulate emotions often more than other types of documents. As we want our system to be easily usable, we selected sensors embedded in a wristband and an eye tracker. Emotions can be described by two dimensions called emotional valence and arousal. As a first step, we propose in this paper to estimate the emo- tional arousal. We analyze the electrodermal activity, blood volume pulse, heart rate, skin temperature and pupil diam- eter of a subject to estimate if the reader feels a high or low arousal while reading. Our experiment shows that for some participants, the arousal can be estimated accurately."}},{"key":"Sanches2016_SCAP","id":"1534","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Charles Lima Sanches","en":"Charles Lima Sanches"},{"ja":"Olivier Augereau","en":"Olivier Augereau"},{"ja":"Koichi Kise","en":"Koichi Kise"}],"title":{"ja":"Manga content analysis using physiological signals","en":"Manga content analysis using physiological signals"},"journal":{"ja":"Proceedings of the 1st International Workshop on coMics ANalysis, Processing and Understanding","en":"Proceedings of the 1st International Workshop on coMics ANalysis, Processing and Understanding"},"book_title":"Proceedings of the 1st International Workshop on coMics ANalysis, Processing and Understanding","series":"MANPU '16","pages":"6:1--6:6","num_pages":"6","year":2016,"month":12,"publisher":"ACM","address":"New York, NY, USA","location":"Cancun, Mexico","review":"reviewed","fields":[],"doi":"10.1145/3011549.3011555","abstract":{"ja":"Recently, the physiological signals have been analyzed more and more, especially in the context of everyday life activities such as watching video or looking at pictures. Tracking these signals gives access to the mental state of the user (interest, tiredness, stress) but also to his emotions (sadness, fright, happiness). The analysis of the reader's physiological signals during reading can provide a better understanding of the reader's feelings but also a better understanding of the documents. Our main research direction is to find the relationship between a change in the reader's physiological signal and the content of the reading. As a first step, we investigate whether it is possible to distinguish a manga (Japanese comic book) from another by analyzing the physiological signals of the reader. We use 3 different manga genres (horror, romance, comedy) and try to predict which one is read by analyzing the features extracted from the physiological signals of the reader. Our method uses the blood volume pulse, the electrodermal activity and the skin temperature of the reader while reading. We show that by using these physiological signals with a support vector machine we can retrieve which manga has been read with a 90% average accuracy.","en":"Recently, the physiological signals have been analyzed more and more, especially in the context of everyday life activities such as watching video or looking at pictures. Tracking these signals gives access to the mental state of the user (interest, tiredness, stress) but also to his emotions (sadness, fright, happiness). The analysis of the reader's physiological signals during reading can provide a better understanding of the reader's feelings but also a better understanding of the documents. Our main research direction is to find the relationship between a change in the reader's physiological signal and the content of the reading. As a first step, we investigate whether it is possible to distinguish a manga (Japanese comic book) from another by analyzing the physiological signals of the reader. We use 3 different manga genres (horror, romance, comedy) and try to predict which one is read by analyzing the features extracted from the physiological signals of the reader. Our method uses the blood volume pulse, the electrodermal activity and the skin temperature of the reader while reading. We show that by using these physiological signals with a support vector machine we can retrieve which manga has been read with a 90% average accuracy."}},{"key":"Iwamura2016_ECCV","id":"1528","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Takahiro Matsuda","en":""},{"ja":"Naoyuki Morimoto","en":""},{"ja":"Hitomi Sato","en":""},{"ja":"Yuki Ikeda","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Downtown Osaka Scene Text Dataset"},"journal":{"ja":"Proc. 2nd International Workshop on Robust Reading (ECCV 2016 Workshops, Part I)"},"volume":"9913","series":"Lecture Notes in Computer Science (LNCS)","pages":"1-16","year":2016,"month":10,"publisher":"Springer","location":"Amsterdam, Netherlands","review":"reviewed","presen_type":"oral","fields":["pattern-recognition"],"doi":"10.1007/978-3-319-46604-0_32","abstract":{"ja":"This paper presents a new scene text dataset named Downtown Osaka Scene Text Dataset (in short, DOST dataset). The dataset consists of sequential images captured in shopping streets in downtown Osaka with an omnidirectional camera. Unlike most of existing datasets consisting of scene images intentionally captured, DOST dataset consists of uncontrolled scene images; use of an omnidirectional camera enabled us to capture videos (sequential images) of whole scenes surrounding the camera. Since the dataset preserved the real scenes containing texts as they were, in other words, they are scene texts in the wild. DOST dataset contained 32,147 manually ground truthed sequential images. They contained 935,601 text regions consisting of 797,919 legible and 137,682 illegible. The legible regions contained 2,808,340 characters. The dataset is evaluated using two existing scene text detection methods and one powerful commercial end-to-end scene text recognition method to know the difficulty and quality in comparison with existing datasets."}},{"key":"Fan2016","id":"1512","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Wei Fan","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Automatic Character Labeling for Camera Captured Document Images"},"journal":{"ja":"Proc. 2016 The International Conference on Image Processing (ICIP 2016)"},"pages":"3284-3288","year":2016,"month":9,"location":"Phoenix, Arizona, USA","review":"reviewed","fields":["document-image-analysis"],"doi":"10.1109/ICIP.2016.7532967","abstract":{"ja":"Character groundtruth for camera captured documents is crucial for training and evaluating advanced OCR algorithms. Manually generating character level groundtruth is a time consuming and costly process. This paper proposes a robust groundtruth generation method based on document retrieval and image registration for camera captured documents. We use an elastic non-rigid alignment method to fit the captured document image which relaxes the flat paper assumption made by conventional solutions. The proposed method allows building very large scale labeled camera captured documents dataset, without any human intervention. We construct a large labeled dataset consisting of 1 million camera captured Chinese character images. Evaluation of samples generated by our approach showed that 99.99% of the images were correctly labeled, even with different distortions specific to cameras such as blur, specularity and perspective distortion."}},{"key":"Augereau2016_ACM","id":"1533","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Olivier Augereau","en":""},{"ja":"Kai Kunze","en":""},{"ja":"Hiroki Fujiyoshi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Estimation of english skill with a mobile eye tracker"},"journal":{"ja":"Proceedings of the 2016 ACM International Joint Conference on Pervasive and Ubiquitous Computing"},"pages":"1777-1781","year":2016,"month":9,"review":"reviewed","fields":["learning-analytics","reading-life-log","sensing"]},{"key":"Ishimaru2016_ACM_Reading","id":"1692","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Shoya Ishimaru","en":"Shoya Ishimaru"},{"ja":"Tilman Dingler","en":"Tilman Dingler"},{"ja":"Kai Kunze","en":"Kai Kunze"},{"ja":"Koichi Kise","en":"Koichi Kise"},{"ja":"Andreas Dengel","en":"Andreas Dengel"}],"title":{"ja":"Reading interventions: tracking reading state and designing interventions","en":"Reading interventions: tracking reading state and designing interventions"},"journal":{"ja":"Proceedings of the 2016 ACM International Joint Conference on Pervasive and Ubiquitous Computing: Adjunct","en":"Proceedings of the 2016 ACM International Joint Conference on Pervasive and Ubiquitous Computing: Adjunct"},"pages":"1759-1764","year":2016,"month":9,"fields":["reading-life-log"],"doi":"10.1145/2968219.2968271","abstract":{"ja":"The quantified-self movement has brought us a multitude of tracking devices mostly focused on physical activities. But healthy user habits go beyond the physical realm, such as cognitive activities, learning and reading. For installing healthy reading habits we built a wordometer using smart eyewear for tracking online and offline reading activities throughout the day. Besides providing reading statistics at the end of the day we propose a number of real-time interventions to support reading in-situ: adjustable text that takes into account users' current cognitive state, such as attention span, visual fatigue and comprehension levels. In this paper we report insights from our in-the-wild reading tracker and explore various intervention techniques with the goal of supporting user attention, focus, and comprehension in reading tasks.","en":"The quantified-self movement has brought us a multitude of tracking devices mostly focused on physical activities. But healthy user habits go beyond the physical realm, such as cognitive activities, learning and reading. For installing healthy reading habits we built a wordometer using smart eyewear for tracking online and offline reading activities throughout the day. Besides providing reading statistics at the end of the day we propose a number of real-time interventions to support reading in-situ: adjustable text that takes into account users' current cognitive state, such as attention span, visual fatigue and comprehension levels. In this paper we report insights from our in-the-wild reading tracker and explore various intervention techniques with the goal of supporting user attention, focus, and comprehension in reading tasks."}},{"key":"Rigaud2016","id":"1566","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"","en":"Christophe Rigaud"},{"ja":"","en":"Thanh-Nam Le"},{"ja":"","en":"J.-C. Burie"},{"ja":"","en":"J.-M. Ogier"},{"ja":"","en":"Shoya Ishimaru"},{"ja":"","en":"Motoi Iwata"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Semi-Automatic Text and Graphics Extraction of Manga Using Eye Tracking Information"},"journal":{"en":"Proc. of 12th IAPR international workshop on Document Analysis Systems (DAS2016)"},"year":2016,"month":4,"location":"Santorini, Greece","review":"reviewed","fields":["document-image-analysis"],"doi":"10.1109/DAS.2016.72","url":"http://ieeexplore.ieee.org/document/7490104/","abstract":{"en":"The popularity of storing, distributing and reading comic books electronically has made the task of comics analysis an interesting research problem. Different work have been carried out aiming at understanding their layout structure and the graphic content. However the results are still far from universally applicable, largely due to the huge variety in expression styles and page arrangement, especially in manga (Japanese comics). In this paper, we propose a comic image analysis approach using eye-tracking data recorded during manga reading sessions. As humans are extremely capable of interpreting the structured drawing content, and show different reading behaviors based on the nature of the content, their eye movements follow distinguishable patterns over text or graphic regions. Therefore, eye gaze data can add rich information to the understanding of the manga content. Experimental results show that the fixations and saccades indeed form consistent patterns among readers, and can be used for manga textual and graphical analysis."}},{"key":"Sato2016","id":"1505","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Tomokazu Sato","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Kitahiro Kaneda","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Fast and Memory Saving Instance Search with Approximate Reverse Nearest Neighbor Search Using Reverse Lookup"},"journal":{"ja":"Proc. Second IEEE International Conference on Multimedia Big Data (BigMM2016)"},"pages":"326-333","year":2016,"month":4,"location":"Taipei, Taiwan","review":"reviewed","presen_type":"oral","fields":["nearest-neighbor-search"],"doi":"10.1109/BigMM.2016.76","abstract":{"ja":"Recently more and more videos have been shared\nthrough the websites such as youtube.com. In order to utilize\nthem efficiently, instance search (INS) techniques which find a\nspecific person, object and place from a video database without\nmetadata has been desired. It is known that the BM25 scoring\nmethod is a powerful tool for the INS task. It is, however, also\nknown that it requires a time consuming process. It has been\npointed out that the time consuming process is equivalent to the\nbichromatic reverse nearest neighbor (BRNN) search problem\nand a method to approximate it has been proposed. However,\nthe algorithm needs a huge reference table which causes large\nmemory usage and computational complexity. In this paper, we\npropose a more efficient way to search BRNNs using a reverse\nlookup structure. An experimental result using the database of\nTRECVID 2012 INS task showed that the proposed method\nwas 2.8 times faster with 10% less memory usage than the\nconventional method."}},{"key":"Oguma2016_DAS","id":"1499","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Yusuke Oguma","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Camera-Based System for User Friendly Annotation of Documents"},"journal":{"ja":"Proceedings of the IAPR International Workshop on Document Analysis Systems (DAS) 2016"},"pages":"434-439","year":2016,"month":4,"location":"Santorini, Greece","review":"reviewed","presen_type":"poster","fields":["document-image-analysis"]},{"key":"Ishimaru2016_ACM_Wordometer","id":"1667","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Shoya Ishimaru","en":"Shoya Ishimaru"},{"ja":"Kai Kunze","en":"Kai Kunze"},{"ja":"Koichi Kise","en":"Koichi Kise"},{"ja":"Andreas Dengel","en":"Andreas Dengel"}],"title":{"ja":"The Wordometer 2.0: Estimating the Number of Words You Read in Real Life using Commercial EOG Glasses","en":"The Wordometer 2.0: Estimating the Number of Words You Read in Real Life using Commercial EOG Glasses"},"journal":{"ja":"Proceedings of the 2016 ACM International Joint Conference on Pervasive and Ubiquitous Computing","en":"Proceedings of the 2016 ACM International Joint Conference on Pervasive and Ubiquitous Computing"},"pages":"293-296","year":2016,"review":"reviewed","fields":["learning-analytics","sensing"]},{"key":"Miki2016_PS","id":"1529","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"三木 啓輔","en":"Keisuke Miki"},{"ja":"内海 ゆづ子","en":"Yuzuko Utsumi"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"アピアランスと平滑性に基づいた植物の 3 次元復元結果の補完","en":"3D reconstruction of a plant based on\u000b appearance and surface smoothness"},"journal":{"ja":"第19回画像の認識・理解シンポジウム予稿集"},"presen_id":"PS2-40","year":2016,"month":8,"review":"not_reviewed","fields":[]},{"key":"Iwamura2016_AAC","id":"1527","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":""},{"ja":"宮田 武嗣","en":""},{"ja":"程 征","en":""},{"ja":"田井中 渓志","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"視覚障害者のための環境文字情報提示システムの検討"},"journal":{"ja":"情報処理学会 研究報告アクセシビリティ（AAC）"},"volume":"2016-AAC-1","number":"7","year":2016,"month":7,"review":"not_reviewed","fields":["pattern-recognition","visually-impaired-assistance"],"url":"http://id.nii.ac.jp/1001/00169969/","note":{"ja":"ここに掲載した著作物の利用に関する注意 本著作物の著作権は情報処理学会に帰属します。本著作物は著作権者である情報処理学会の許可のもとに掲載するものです。ご利用に当たっては「著作権法」ならびに「情報処理学会倫理綱領」に従うことをお願いいたします。","en":"Notice for the use of this material The copyright of this material is retained by the Information Processing Society of Japan (IPSJ). This material is published on this web site with the agreement of the author (s) and the IPSJ. Please be complied with Copyright Law of Japan and the Code of Ethics of the IPSJ if any users wish to reproduce, make derivative work, distribute or make available to the public any part or whole thereof.\nAll Rights Reserved, Copyright (C) Information Processing Society of Japan.\nComments are welcome. Mail to address editj＠ipsj.or.jp, please."},"abstract":{"ja":"本稿では，視覚障害者が環境中に存在する文字情報を利用できるようにする情報提示システムを検討する．文字認識技術を利用して視覚障害者に文字情報を提示する試みは過去にもあるが，いずれも文字が存在する場所，あるいはその範囲がわかっている場合に使用できるように設計されている．視覚に障害がある人の中には失語症のように，文字を指し示すことのできる人もいるが，全盲や弱視の人にとって，文字を自力で発見することは必ずしも容易でないと考えられる．そのため，従来からあるような文字の読み上げ機能や，文字の場所を示す機能に加えて，文字を発見することに工夫する．具体的には，利用者の前方だけでなく，後や横を含めた周囲の文字を小型の全方位カメラを利用して発見する．これにより利用者が自ら文字を発見する必要はなくなるものの，より広範囲をカバーすることが必要になるため，従来からある文字の場所を示す方法が使用できなくなる．そのため，立体音響と頭の向きを実時間で追従するインタフェースを組み合わせることで場所を示すことを検討する．前報 [1] では，オープン型のヘッドホンを使用したシステムを提案したが，オープン型とはいえヘッドホンで耳を塞いでしまうと音に敏感な視覚障害者の貴重な情報源を塞いでしまうのではないかというコメントを受けて，本稿では耳を塞がない骨伝導イヤホンを用いた場合の性能を評価する．"}},{"key":"Hoshika2016_PRMU","id":"1523","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"星加 健介","en":""},{"ja":"石丸 翔也","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"リーディングライフログを用いた日常的な読書行動の計量"},"book_title":"電子情報通信学会技術研究報告, パターン認識メディア理解研究会","volume":"116","number":"38","presen_id":"PRMU2016-19","pages":"101-106","year":2016,"month":5,"location":"名古屋大学","review":"not_reviewed","presen_type":"oral","fields":[],"abstract":{"ja":"本稿では，日常的な読書行動を検出するために，視点情報と深層学習を用いた読書判定方法を試みる．人の視点を計測できるデバイスを使用し，読むという行為を記録したものをリーディングライフログという．読書行動を記録できると，何を読んだのか，どのくらい読んだのかという読書傾向の変化を知ることができる．現在のリーディングライフログの研究では，実験的な環境で小規模なものが多い．その理由は，精度の高いアイトラッカは高価であり、アイトラッカによってはバッテリの問題があるからである．上記の理由から，日常的な読書行動を大規模に取得する研究は少ない．そこで本稿では，大規模に記録を行うために，廉価なアイトラッカであるTobii EyeX Controllerに注目した．Tobii EyeX Controllerを用いることで，ディスプレイ上の人の視点情報を記録することができる．アイトラッカから取得した視点情報を用いて，人の日常的な読書行動を計量するためには，人が読書中か判定する必要がある．ディスプレイ上の様々なレイアウトの文字において，読書判定を行うために，深層学習を用いる．実験では，被験者5人において，被験者依存で学習し，読書判定を行った．その結果，被験者5人の平均適合率は0.73であった．"}},{"key":"Mahad2016_CVIM","id":"1526","type":"domestic_conf","language":"EN","show":"show","authors":[{"ja":"","en":"Fairuz Safwan Mahad"},{"ja":"","en":"Masakazu Iwamura"},{"ja":"","en":"Yuzuko Utsumi"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Denser Feature Correspondences for 3D Reconstruction"},"journal":{"en":"IPSJ SIG-CVIM: Computer Vision and Image Media"},"volume":"2016-CVIM-202","number":"3","pages":"1-5","year":2016,"month":5,"location":"Ritsumeikan University","review":"not_reviewed","presen_type":"oral_and_poster","fields":["3d-reconstruction"],"url":"http://id.nii.ac.jp/1001/00159114/","note":{"ja":"ここに掲載した著作物の利用に関する注意 本著作物の著作権は情報処理学会に帰属します。本著作物は著作権者である情報処理学会の許可のもとに掲載するものです。ご利用に当たっては「著作権法」ならびに「情報処理学会倫理綱領」に従うことをお願いいたします。","en":"Notice for the use of this material The copyright of this material is retained by the Information Processing Society of Japan (IPSJ). This material is published on this web site with the agreement of the author (s) and the IPSJ. Please be complied with Copyright Law of Japan and the Code of Ethics of the IPSJ if any users wish to reproduce, make derivative work, distribute or make available to the public any part or whole thereof.\nAll Rights Reserved, Copyright (C) Information Processing Society of Japan.\nComments are welcome. Mail to address editj＠ipsj.or.jp, please."},"abstract":{"en":"The 3D reconstruction of an object is heavily dependent on the amount of texture on the surface of the object and the correspondences acquired between images from different viewpoints. Objects with lack of texture affects the reconstructed 3D model degrading its quality and completeness. This occurs due to the SIFTs and Harris's incapability of extracting sufficient features from regions with less texture thus resulting in a lack of correspondences in such regions. Gaps occurs in the reconstructed 3D model due to the lack of correspondences. The paper aims to obtain more correspondences through the implementation of convolution in conjunction with a set of randomized kernels. The use of randomized kernels changes the structure of the images in different ways allowing SIFT to extract features in less textured regions which could not be realized conventionally. Experimental results showed that the proposed method is able to fill up more gaps improving the completeness of the 3D model of less textured objects."}},{"key":"Miki2016_CVIM","id":"1520","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"三木啓輔","en":"Keisuke Miki"},{"ja":"内海ゆづ子","en":"Yuzuko Utsumi"},{"ja":"岩村雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬浩一","en":"Koichi Kise"}],"title":{"ja":"アピアランスに基づいた植物の3次元復元結果の補完"},"journal":{"ja":"研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2016-CVIM-202","number":"29","pages":"1-7","year":2016,"month":5,"location":"立命館大学・大阪いばらきキャンパス","review":"not_reviewed","presen_type":"poster","fields":["3d-reconstruction"],"url":"http://id.nii.ac.jp/1001/00159140/","note":{"ja":"ここに掲載した著作物の利用に関する注意 本著作物の著作権は情報処理学会に帰属します。本著作物は著作権者である情報処理学会の許可のもとに掲載するものです。ご利用に当たっては「著作権法」ならびに「情報処理学会倫理綱領」に従うことをお願いいたします。","en":"Notice for the use of this material The copyright of this material is retained by the Information Processing Society of Japan (IPSJ). This material is published on this web site with the agreement of the author (s) and the IPSJ. Please be complied with Copyright Law of Japan and the Code of Ethics of the IPSJ if any users wish to reproduce, make derivative work, distribute or make available to the public any part or whole thereof.\nAll Rights Reserved, Copyright (C) Information Processing Society of Japan.\nComments are welcome. Mail to address editj＠ipsj.or.jp, please."},"abstract":{"ja":"近年，植物フェノタイピングの自動化を目的とした，植物の 3 次元復元に関する研究に注目が集まっている．3 次元復元の方法として，カメラで撮影した画像を用いる方法が注目されている．しかし，画像を用いて植物を 3 次元復元した場合，主に葉の部分などに欠落が生じる．そこで，本研究では，アピアランスに基いて植物の 3 次元復元結果を補完する手法を提案する．提案手法では，復元されなかった箇所の奥行きを画像の色情報をもとに推定し，補完を行う．実験の結果，欠落箇所を補完することができた．"}},{"key":"Nakajima2016_CVIM","id":"1524","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"中嶌 一樹","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"読書行動の検出における有効な視線特徴量の選定"},"journal":{"ja":"研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2016-CVIM-202","number":"28","pages":"1-6","year":2016,"month":5,"location":"立命館大学・大阪いばらきキャンパス","review":"not_reviewed","fields":["reading-life-log","pattern-recognition"],"url":"http://id.nii.ac.jp/1001/00159139/","note":{"ja":"ここに掲載した著作物の利用に関する注意 本著作物の著作権は情報処理学会に帰属します。本著作物は著作権者である情報処理学会の許可のもとに掲載するものです。ご利用に当たっては「著作権法」ならびに「情報処理学会倫理綱領」に従うことをお願いいたします。","en":"Notice for the use of this material The copyright of this material is retained by the Information Processing Society of Japan (IPSJ). This material is published on this web site with the agreement of the author (s) and the IPSJ. Please be complied with Copyright Law of Japan and the Code of Ethics of the IPSJ if any users wish to reproduce, make derivative work, distribute or make available to the public any part or whole thereof.\nAll Rights Reserved, Copyright (C) Information Processing Society of Japan.\nComments are welcome. Mail to address editj＠ipsj.or.jp, please."}},{"key":"Fujiyoshi2016","id":"1496","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"藤好 宏樹","en":"Hiroki Fujiyoshi"},{"ja":"石丸 翔也","en":"Shoya Ishimaru"},{"ja":"Olivier Augereau","en":"Plivier Augereau"},{"ja":"黄瀬 浩一","en":"Koichi kise"}],"title":{"ja":"視点情報を用いた英語習熟度推定法の実験的検討","en":"Experimental investigation for estimation method of English proficiency using eye gaze information"},"book_title":"電子情報通信学会技術研究報告","volume":"115","number":"517","presen_id":"PRMU2015-195","pages":"185-190","year":2016,"month":3,"location":"東京都江東区","review":"not_reviewed","presen_type":"oral","fields":["pattern-recognition"],"abstract":{"ja":"本稿では，アイトラッカで取得した英文問題解答時の視点情報を用いて，英語習熟度を推定する手法を提\n案する．一般に，英語習熟度を測るテストとしてTOEIC が挙げられる．しかし，TOEIC には結果が出るまでに時間\nを要することや，問題数が多く受験者に負担が大きいという課題がある．そこで，TOEIC よりも少ない問題数で手軽\nに英語習熟度を測る事ができれば有益であると考え，我々は問題解答時の視点情報を用いてTOEIC スコアを推定す\nる手法を提案した．しかし，この従来法には，推定に用いる文書数が10 題と多く，ユーザの負担が依然として大きい\nことが課題となっている．本研究では，従来法で用いたものに比べて高精度なアイトラッカを使用してより多くの被\n験者，文書数のデータを収集した．加えて，新しい特徴量を用いて推定することやテストデータと学習データを変化\nさせた際の推定結果を調査するなど，英語習熟度推定法について検討を行う．結果として英文問題3 題を用いて被験\n者16 人の英語習熟度を推定したところ，TOEIC スコアを平均絶対誤差32 点で推定することができた."}},{"key":"Matsubara2016_PRMU","id":"1497","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"松原 瑞氣","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"岩田 基","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"視点情報を用いた電子コミックの表示方法"},"book_title":"電子情報通信学会技術研究報告","volume":"115","number":"517","presen_id":"PRMU2015-188","pages":"143-148","year":2016,"month":3,"location":"東京都江東区","review":"not_reviewed","fields":[],"abstract":{"ja":"近年,多くの人がスマートフォンなどの小型端末を用いて,電子コミックを読んでいる.小型端末の画面 に電子コミックの 1 ページ全体を表示すると,小さな文字が読めない場合や,絵が見えにくい場合がある.本稿では, 事前に記録された電子コミックを読んでいる際の視点を用いて,電子コミックを拡大表示する手法を提案する.提案 手法により,小型端末の画面でも見やすい縮尺で電子コミックを表示することが可能となる.また,本稿では被験者 に提案手法を用いて電子コミックを読んでもらうことで,その使いやすさを調査する.その結果,提案手法により, 文字や絵が見やすい大きさでの表示が可能であることが分かった.また,今後の課題として,拡大率を自動で最適な 値にする機能が必要であることが分かった."}},{"key":"Oguma2016_PRMU","id":"1487","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"小熊 勇佑","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"多様な文書にも対応可能なカメラベース文書アノーテーション付与・共有"},"book_title":"電子情報通信学会技術研究報告","volume":"115","number":"517","presen_id":"PRMU2015-187","pages":"137-142","year":2016,"month":3,"location":"東京都江東区","review":"not_reviewed","presen_type":"oral","fields":["document-image-retrieval","document-image-analysis"],"abstract":{"ja":"スマートフォンのカメラを用いた拡張現実による文書アノテーションシステムを提案する．\n本システムの特徴は，紙文書と電子文書を垣根無く扱える点である．これは，\n両者を文書画像として扱い，アノーテーションを拡張現実として表示する\nことによって可能となる．アノーテーションと文書を対応付けて拡張現実として\n表示するためには画像検索が必須となる．この際，\nテキスト，図表，写真などが混在する多様な文書にいかに対応するかが\n重要となる．\nすでに，テキスト部分のみを対象とした高速な検索法や，\n図表・写真などにも対応可能な一般的な検索法は存在\nするものの，前者は図表・写真を対象にできない，後者は速度に問題があるなど，\n多様な文書に対応できていない．そこで本稿では，両者を組み合わせる\nことによって，高速性をあまり犠牲にぜずに多様な文書に対応する手法を提案する．\n加えて，付加するアノーテーションをスタンプ形式としてスマートフォンで容易に使用可能と\nしたり，印刷物に手書きで加えられたアノーテーションをディジタルデータとして\n分離して電子文書に付与する方式についても述べる．"}},{"key":"Okoso2016","id":"1495","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"大社 綾乃","en":"Ayano Okoso"},{"ja":"石丸 翔也","en":"Syoya Ishimaru"},{"ja":"Olivier Augereau","en":"Olivier Augereau"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"視点情報を用いた主観的高難易度単語の推定","en":"Estimation of subjectively difficult words on documents by using gaze data"},"book_title":"電子情報通信学会技術研究報告","volume":"115","number":"517","presen_id":"PRMU2015-189","pages":"149-154","year":2016,"month":3,"location":"東京都江東区","review":"not_reviewed","presen_type":"oral","fields":[],"abstract":{"ja":"本稿では，アイトラッカから取得した視点情報より，Deep Learning を用いた，文書上の主観的高難易度\n単語の推定手法を提案する．「主観的高難易度単語」とは，文書を読んだ際にユーザが難しいと感じた単語を表す．文\n書を読むだけで，主観的高難易度単語を推定することが出来れば，ユーザはどの単語を学習しなければならないかを\n簡単に知ることが出来る．文書を読んでいる際に，難しいと感じた単語に印をつける手間が省け，読むリズムが乱れ\nないという利点もある．そのため，学習補助に大変有効であると言える．一般的に，難しい単語を目にすると，読む\n速度が遅くなったり，何度も読み返すといったような視点パターンがみられる傾向がある．しかし，難しい単語の意\n味を文脈から判断できる場合，読み飛ばし，視点パターンに特徴が見られないことがある．そこで，提案手法では視\n点情報に加え，単語の出現頻度情報を用いる．まず，各単語を中心に前後２〜３単語を読んでいる際の視点パターン\nを抽出し，視点情報による入力データを得る．そして，入力データに単語の一般的な出現頻度情報を付加し，Deep\nLearning を用いて学習することで，より高精度な主観的高難易度単語の推定を目指す．提案手法の有効性を示すため，\n25 名の視線データを記録し学習し，主観的高難易度単語の推定精度を検証した．検証の結果，推定精度は，再現率\n90:8%，適合率34:4%，F 値49:2% であった．"}},{"key":"Mori2016_CVIM","id":"1491","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"森 俊介","en":""},{"ja":"岩村雅一","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"黄瀬 浩一","en":""},{"ja":"田上 拓弥","en":""},{"ja":"槇原 靖","en":""},{"ja":"村松 大吾","en":""},{"ja":"八木 康史","en":""}],"title":{"ja":"大規模歩容モデルの高速検索に基づく歩容認証のための人物領域抽出"},"journal":{"ja":"情報処理学会研究報告コンピュータビジョンとイメージメディア(CVIM)"},"book_title":"情報処理学会研究報告","volume":"57","number":"11","presen_id":"CVIM2016-11","year":2016,"month":3,"review":"not_reviewed","presen_type":"oral","fields":[],"note":{"ja":"ここに掲載した著作物の利用に関する注意 本著作物の著作権は情報処理学会に帰属します。本著作物は著作権者である情報処理学会の許可のもとに掲載するものです。ご利用に当たっては「著作権法」ならびに「情報処理学会倫理綱領」に従うことをお願いいたします。","en":"Notice for the use of this material The copyright of this material is retained by the Information Processing Society of Japan (IPSJ). This material is published on this web site with the agreement of the author (s) and the IPSJ. Please be complied with Copyright Law of Japan and the Code of Ethics of the IPSJ if any users wish to reproduce, make derivative work, distribute or make available to the public any part or whole thereof.\nAll Rights Reserved, Copyright (C) Information Processing Society of Japan.\nComments are welcome. Mail to address editj＠ipsj.or.jp, please."},"abstract":{"ja":"歩容認証では，人物領域のシルエットに基づいて抽出される特徴を用いる手法が主流である．\nこの手法では，人物領域の抽出精度が一般に性能を左右するため，人物領域を高精度に抽出できる手法が求められる．\n人物領域抽出において，歩容の事前知識である標準歩容モデルを用いることで抽出精度が向上することが報告されている．\n多くの標準歩容モデルを用いると，認証精度が向上すると考えられるが，標準歩容モデルが増えると処理時間が増加するため，大規模な標準歩容モデルを用いることは難しい．\nこの問題を解決するため，本研究では近似最近傍探索を利用して標準歩容モデル選択を高速化する手法を提案する．\nこれにより，処理時間が大きく削減できるため，従来よりも大規模な標準歩容モデルを用いることが容易になる．\nそこで，認証精度の向上のために，大規模な標準歩容モデルを用いて人物領域を高精度に抽出できるか検証する．\n% これにより大規模な標準歩容モデルを用いた人物領域の抽出が実時間で可能となり，また人物領域の抽出精度が向上して認証精度が向上する．\n実験の結果，従来の認証精度を維持したまま標準歩容モデルの選択時間を82%削減でき，人物領域抽出に大規模な標準歩容モデルを用いることで認証精度が向上することが確認できた．"}},{"key":"Miyata2016","id":"1494","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"宮田 武嗣","en":"Taketsugu Miyata"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"立体音響で教える全方位単語感知システム","en":"Omnidirectional Word Sensing System Guiding with Stereophonic Voice"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"115","number":"517","presen_id":"PRMU2015-194","pages":"179-184","year":2016,"month":3,"location":"東京都江東区","review":"not_reviewed","presen_type":"oral","fields":["pattern-recognition","visually-impaired-assistance"],"abstract":{"ja":"私たちは身の回りにある文字情報を見て必要な情報を取り入れるが，視覚障害者はそれができない．その ような人が視覚の代わりに触覚や聴覚を使って情報を取り入れられるような補助ツールとして点字や音声案内がある． しかし点字は設置している箇所に辿り着けなければ利用できないことや，音声案内はむやみに設置すると健常者に とっては騒音となる問題がある．それに対してこれまで様々なウェアラブルインタフェースが提案されてきたが，ど こに文字情報があるか分からなければ使用できないことや，正面以外の方向にある文字情報を認識できないといった 問題がある．そこで全方位から文字情報を探し出し，何という文字情報がどこにあるか声で教えるウェアラブルイン タフェースを提案する．このインタフェースでは，実際に文字情報が位置する方向から声が聞こえるように，立体音 響を用いて文字情報を読み上げる．本稿では，文字情報の位置を教えるために立体音響を用いることが適切がどうか 確かめるべく，評価実験を行った．"}},{"key":"Ogawa2016","id":"1500","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"小川 広晃","en":"Hiroaki Ogawa"},{"ja":"岩田 基","en":"Motoi Iwata"},{"ja":"黄瀬 浩一","en":"Koichi kise"}],"title":{"ja":"画像検索を用いて位置検出用マーカのない印刷物から透かしを抽出できる電子透かし法","en":"Digital Watermarking Method Using Image Retrieval to Printed Images without Finder Patterns and Alignment Pattern"},"journal":{"ja":"2016年 暗号と情報セキュリティシンポジウム予稿集","en":"Proc. of 2016 Symposium on Cryptography and Information Security (SCIS2016)"},"presen_id":"3C1-2","num_pages":"7","year":2016,"month":1,"review":"not_reviewed","fields":[],"abstract":{"ja":"電子透かしを用いて，印刷物に人間には認識できないように情報を埋め込み，それをスマートフォンで撮影して情報を取り出す手法について検討する．SCIS2013 に投稿した手法を基に手法の改良を行った．従来の電子透かし法では位置検出のためにマーカや，枠線をつける必要があった．本来，電子透かしの利点の一つは，見た目を損なわないという点にあるのだが，マーカや，枠線をつけることはその利点と相反している．本稿では，この問題点を解決するために，位置検出の手法を改善し，画像に何も付与せずに位置検出し，透かしから情報を取り出す手法を提案する．画像に何も付与せずに位置検\n出を行うために，画像検索を用いて撮影画像内にのどの位置に透かし入り画像が存在するのかを検出し，位置検出を行う．画像検索は計算コストが高いため，位置検出はスマートフォン上でなくサーバ上で行っている．位置検出法を変更した結果，従来法と比べ処理時間について従来法が0.34 秒だったのに対して0.9 秒を要するようになったものの，同等の抽出率を保持できた．"}},{"key":"Mizushima2016_Symp","id":"1490","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"水島 尚良","en":""},{"ja":"岩田 基","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"スマートフォンを用いた透かし入り動画から透かし領域を推定する電子透かし法の性能評価"},"journal":{"ja":"2016年暗号と情報セキュリティシンポジウム予稿集"},"presen_id":"3C1-4","num_pages":"7","year":2016,"month":1,"review":"not_reviewed","fields":[],"abstract":{"ja":"近年，スマートフォンやデジタルサイネージ上での広告などが普及している．\n 電子透かしを用いて動画に対し付加情報を埋め込んで透かし入り動画を作成し，スマートフォンで再撮影して情報を取り出せる既存の手法に，岩田らの手法がある．\n 岩田らは透かし入り動画にスマートフォンをかざすことによって関連情報にアクセスできるようにすることを目的としており，\niPhone5sで透かし入り動画を撮影して透かしを抽出し，性能を評価している．\n 岩田らの手法では今後の課題の1つとして，スマートフォンでの撮影時の手振れの影響の軽減が挙げられていた．\nこれに対し，iPhone6sPlusには光学手振れ補正機能が搭載されており，撮影時の手振れの影響の軽減が見込まれる．\nまた，岩田らの性能評価ではノートPC上での映像を対象としていたため，より実用に近い，70インチのサイネージを実験に使用する．\n 以上より，本稿では，サイネージ上で再生した透かし入り動画をiPhone6sPlusで撮影して透かしを抽出し，性能を評価する．"}},{"key":"Kitano2016","id":"1779","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"北野 豊","en":""}],"title":{"ja":"解像度低下と表情変化に頑健な大規模顔認識","en":"Expression and resolution invariant face recognition on a large scale database"},"year":2016,"month":3,"fields":[]},{"key":"Iwata2015_J","id":"1467","type":"journal_paper","language":"JA","show":"show_except_datafile","authors":[{"ja":"岩田 基","en":"Motoi Iwata"},{"ja":"江口 健太","en":"Kenta Eguchi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"均等色空間からRGB表色系への変換に起因する抽出誤りを防止した電子透かし法","en":"Digital Watermarking Method Preventing Extraction Error Caused by Transformation from Uniform Color Space to RGB Color Space"},"journal":{"ja":"日本画像学会誌","en":"Journal of the Imaging Society of Japan"},"volume":"54","number":"4","pages":"284-291","year":2015,"month":8,"review":"reviewed","fields":[],"abstract":{"ja":"均等色空間とは，人の知覚的な色の違いと均等色空間での画素値の違いが一致するという特性を持つ色空間である．そのため，画素値の変更量と人の知覚との関係を数値化できることが大きな利点となる電子透かしの分野において，注目されてきた．しかしながら，一般的な画像フォーマットであるRGB表色系と均等色空間は非線形な式によって変換される上に，RGB表色系は整数値で表現されるため丸め誤差が生じ，透かしを正しく抽出できない場合があった．この問題に対処した方法として丸め誤差が生じない画素値のみを保持するテーブルを用いた手法がある．この手法では，丸め誤差による誤抽出の防止には成功したものの，透かしの埋め込み処理に要する時間が膨大であることや探索方法を原因とする誤抽出が生じるという問題があった．本論文では，探索方法の改善と表色系変換テーブルの分割により，処理の高速化と誤抽出の完全な防止を実現する．そして，従来の方法に関する問題点や，提案法の性能について，実験により明らかにする．","en":"Uniform color space is a color space matching with human visual system．Therefore many digital watermarking method employed it．However, watermark bits are not always correctly extracted if the RGB values transformed from uniform color space are adjusted to integer values by rounded-up although this procedure is often employed．It is because the transformation from uniform color space to RGB color space is non-linear．The proposed method solves this problem by using color transformation table．Moreover, the proposed method improves processing speed by changing search procedure and dividing color transformation table appropriately．Experimental results clarify the problem of existing methods and the performance of the proposed method．"}},{"key":"Makihara2015","id":"1465","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Yasushi Makihara","en":""},{"ja":"Takuya Tanoue","en":""},{"ja":"Daigo Muramatsu","en":""},{"ja":"Yasushi Yagi","en":""},{"ja":"Syunsuke Mori","en":""},{"ja":"Yuzuko Utsumi","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Individuality-Preserving Silhouette Extraction for Gait Recognition"},"journal":{"ja":"IPSJ Transactions on Computer Vision and Applicaitons (CVA)"},"volume":"7","pages":"74-78","year":2015,"month":7,"review":"reviewed","fields":[],"doi":"10.2197/ipsjtcva.7.74"},{"key":"Horimatsu2015","id":"1430","type":"journal_paper","language":"JA","show":"show_except_datafile","authors":[{"ja":"堀松 晃","en":"Akira Horimatsu"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"特徴点の一意決定性を利用した高速なアフィン不変文字認識","en":"Fast Affine Invariant Character Recognition Using Unique Determination of Feature Points"},"journal":{"ja":"電子情報通信学会論文誌D","en":"Trans. IEICE"},"volume":"J98-D","number":"4","pages":"651-660","year":2015,"month":4,"review":"reviewed","fields":["pattern-recognition"],"doi":"10.14923/transinfj.2014IUP0016","url":"http://search.ieice.org/bin/summary.php?id=j98-d_4_651","abstract":{"ja":"デジタルカメラやスマートフォンで\n情景中の文字を撮影すると，スキャナで撮像した場合とは違い，\n文字画像は様々に劣化する．\nそのため，情景中の文字を認識するためには，文字認識技術がこのような劣化\nに頑健であることが望ましい．\n本論文では，特に射影歪みに注目して，\n射影歪みに対する頑健性と高速性を備えた文字認識手法を提案する．\n提案手法は，文字が連結成分毎に切り出された事を前提とし，\nアフィン歪みに不変な連結成分の照合を高速に実現する．\nこのような照合には，\n特徴点の配置を利用した物体認識法であるGeometric Hashingの考えを\n利用することが考えられるが，単純に適用すると計算量が大きくなってしまう．\nそこで，生成する基底のパターンに制約を設けることで計算量を\n削減する方法を提案する．\n提案手法を用いた実験により，アフィン歪みを受けた文字が\n高速に認識できることを確認した．","en":"Scene text captured with a digital camera or a smartphone suffers from\nvarious degradation.\nThus, it is desired that a character recognition technique is robust\nto the degradation.\nIn this paper, we focus on perspective distortion and propose a\ncharacter recognition method that is fast, invariant to affine\ndistortion and free from layout constraint.\nFor affine invariant matching, geometric hashing is a well-known\nmethod for object recognition.\nHowever, it is computationally too expensive.\nThus, we propose a method to greatly reduce the computational cost."}},{"key":"Kise2015_IEICE","id":"1411","type":"journal_paper","language":"EN","show":"show_except_datafile","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Marcus Liwicki","en":""}],"title":{"ja":"Data Embedding into Characters"},"journal":{"ja":"IEICE Transactions on Information and Systems"},"volume":"E98-D","number":"1","pages":"10-20","year":2015,"month":1,"review":"reviewed","fields":[],"doi":"10.1587/transinf.2014MUI0002","note":{"ja":"Invited paper."}},{"key":"Iwata2015_IAPRPS","id":"1530","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"","en":"Motoi Iwata"},{"ja":"","en":"Eiki Imazu"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Similarity Learning Based on Pool-Based Active Learning for Manga Character Retrieval"},"journal":{"en":"Proc. of the 3rd IAPR Asian Conference on Pattern Recognition"},"presen_id":"PS2-26","num_pages":"6","year":2015,"month":11,"location":"Aloft Kuala Lumpur Sentral Hotel, Kuala Lumpur, Malaysia","review":"reviewed","presen_type":"poster","fields":["information-retrieval","pattern-recognition"],"abstract":{"en":"Manga is Japanese style comics which is one of the most\npopular publications in Japan. There are some e-book services\nfor selling manga. Unfortunately, most of them handle\nmanga only as the set of images. Some digital media use\ntags or text data for retrieval, where the tags and text data\nare produced by handmade input.\nOur final goal is developing a practical content-based\nmanga retrieval system based on automatic tagging of the\ncontents of manga, for example, place, character, expression,\nsituation, and other information readers are interested\nin. At the firsts step of this goal, we focus on effective tagging\nfor manga character retrieval because characters are\nmost important elements in manga. We employ the measurement\n“similarity” to measure the difference between\nline drawings representing characters. Similarity becomes\nhigher when the two line drawings represent a same character.\nSimilarity is obtained as weighted distance of CMRHOG\nfeatures, where the weights are obtained by AdaBoost\nlearning. We use pool-based active learning for selecting\ntraining data efficiently because the training data should be\nlabeled by handmade which is high cost. Experimental results\ndemonstrates the effectiveness of similarity and the selection\nof distinctive training samples based on pool-based\nactive learning."}},{"key":"Ogawa2015_TDFW","id":"1483","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"","en":"Hiroaki Ogawa"},{"ja":"","en":"Minoru Kuribayashi"},{"ja":"","en":"Motoi Iwata"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"DCT-OFDM based watermarking scheme robust against clipping, rotation, and scaling attacks"},"journal":{"en":"14th International Workshop on Digital-forensics and Watermarking"},"presen_id":"46","num_pages":"14","year":2015,"month":10,"review":"reviewed","presen_type":"oral","fields":[],"abstract":{"ja":"The digital watermark is used for copyright protection, and the methods should have the robustness against various attacks such as JPEG compression, clipping, rotation, and scaling attacks. Based on the DCT-OFDM based watermarking method presented at IWIHC2014, we propose a method enhancing the performance in this paper. There are two ideas in the proposed method. One idea is to reduce the computa-\ntional costs required for the synchronization recovery. The other idea is to encode watermark information by error correcting code and to decode it by maximum likelihood algorithm. As the result, our method has high robustness against the IHC evaluation criteria with less degradation of quality and less computational costs."}},{"key":"Oguma2015","id":"1470","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Yusuke Oguma","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Media-Independent Stamp-Based Document Annotation Using DocumentImage Retrieval"},"journal":{"ja":"Proc. of the 1st International Workshop on Visual Recognition and Retrieval for Mixed and Augmented Reality"},"num_pages":"4","year":2015,"month":10,"location":"Fukuoka, Japan","review":"reviewed","presen_type":"oral","fields":["document-image-analysis"],"abstract":{"ja":"In recent years, electronic documents have become popular. One\nof the advantages of electronic documents is that they provide a\nmethod of putting and sharing annotations on documents. However\nthe market size of paper documents is still much larger than\nelectronic documents and we continue to use them. We consider\nthat it is better to have a method of annotation applicable not only\nto electronic documents but also to paper documents. In this paper\nwe propose a method of annotating both electronic and paper\ndocuments by capturing them as images. We use a smartphone as\na device and make the method work in real time. As a way of\nannotation, we propose to use ”stamps”, which are pictorial icons\nrepresenting opinions of readers (like, dislike, difficult, interesting,\netc.). This helps readers to put annotations more easily as compared\nto text-based annotations."}},{"key":"Kunze2015b","id":"1473","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Kai Kunze","en":""},{"ja":"Tilman Dingler","en":""},{"ja":"Niels Henze","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Yoichi Sato","en":""}],"title":{"ja":"2nd Workshop on Ubiquitous Technologies to Augment the Human Mind: Towards the Knowledge Log"},"pages":"1213-1216","year":2015,"month":9,"review":"reviewed","fields":[]},{"key":"Kunze2015a","id":"1471","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Kai Kunze","en":""},{"ja":"Katsutoshi Masai","en":""},{"ja":"Masahiko Inami","en":""},{"ja":"Ömer Sacakli","en":""},{"ja":"Marcus Liwicki","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"Shoya Ishimaru","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Quantifying Reading Habits – Counting How Many Words You Read"},"journal":{"ja":"Proc. of UbiComp 2015"},"pages":"87-96","year":2015,"month":9,"location":"Osaka, Japan","review":"reviewed","presen_type":"oral","fields":["document-image-analysis"]},{"key":"Ishimaru2015_WAHM","id":"1474","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Shoya Ishimaru","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Quantifying the Mental State on the Basis of Physical and Social Activities"},"journal":{"ja":"Proc. of WAHM: 2nd Workshop on Ubiquitous Technologies to Augment the Human Mind: Towards the Knowledge Log"},"pages":"1217-1220","year":2015,"month":9,"location":"Osaka, Japan","review":"reviewed","presen_type":"oral","fields":[],"doi":"10.1145/2800835.2807934","abstract":{"ja":"We demonstrate the idea of \"Thermometer for the Mind\":\nthe mental state estimating system by a user’s activity log\nderived from wearable devices. While it is well known that\nphysical and social activities are correlated with the mental\nstate, our aim is to build an application which tracks daily\nactivities automatically, estimates the mental state and visualizes\nit in an easily understandable way. As a preliminary\nexperiment, we investigated how information about physical\nactivity from a smartphone (step counts) and social activity\nfrom a Web service (Twitter post counts) can be used to\nestimate a user’s mental state. The method is evaluated on\none participant’s 5 months recording. The classification accuracy\nfor 3 classes (mood is low, middle and high) is 60%."}},{"key":"Kudo2015","id":"1475","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Riki Kudo","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"Takuto Rou","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Reading Similarity Measure Based on Comparison of Fixation Sequences"},"journal":{"ja":"Proc. of WAHM: 2nd Workshop on Ubiquitous Technologies to Augment the Human Mind: Towards the Knowledge Log"},"pages":"1221-1226","year":2015,"month":9,"location":"Osaka, Japan","review":"reviewed","presen_type":"oral","fields":[],"doi":"10.1145/2800835.2807935"},{"key":"Sanches2015","id":"1476","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Charles Lima Sanches","en":"Charles Lima Sanches"},{"ja":"Olivier Augereau","en":"Olivier Augereau"},{"ja":"Koichi Kise","en":"Koichi Kise"}],"title":{"ja":"Eye gaze and text line matching for reading analysis","en":"Eye gaze and text line matching for reading analysis"},"journal":{"ja":"Proc. of WAHM: 2nd Workshop on Ubiquitous Technologies to Augment the Human Mind: Towards the Knowledge Log","en":"Proc. of WAHM: 2nd Workshop on Ubiquitous Technologies to Augment the Human Mind: Towards the Knowledge Log"},"book_title":"Adjunct Proceedings of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of the 2015 ACM International Symposium on Wearable Computers","series":"UbiComp/ISWC'15 Adjunct","pages":"1227-1234","num_pages":"7","year":2015,"month":9,"publisher":"ACM","address":"New York, NY, USA","location":"Osaka, Japan","review":"reviewed","presen_type":"oral","fields":[],"doi":"10.1145/2800835.2807936","abstract":{"ja":"Eye tracking data has been widely used to analyze our reading behavior. Usually, experiments are carried out with head fixations or by analyzing eye tracking data in large areas such as paragraphs. But if we want to analyze the eye gaze line by line or word by word with a non invasive apparatus, we have to face the mislocation of the recorded eye gaze. The lack of accuracy involves a difficult analysis of the small eyes movements during reading. This paper proposes a method to match lines of gazes with corresponding text lines, using three different methods. We will show that the Dynamic Time Warping is a promising way to measure similarity between a line of gaze and a text line.","en":"Eye tracking data has been widely used to analyze our reading behavior. Usually, experiments are carried out with head fixations or by analyzing eye tracking data in large areas such as paragraphs. But if we want to analyze the eye gaze line by line or word by word with a non invasive apparatus, we have to face the mislocation of the recorded eye gaze. The lack of accuracy involves a difficult analysis of the small eyes movements during reading. This paper proposes a method to match lines of gazes with corresponding text lines, using three different methods. We will show that the Dynamic Time Warping is a promising way to measure similarity between a line of gaze and a text line."}},{"key":"Kunze2015_UbiComp","id":"1478","type":"intl_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"Kai Kunze","en":""},{"ja":"Yuji Uema","en":""},{"ja":"Katsuma Tanaka","en":""},{"ja":"Shoya Ishimaru","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Masahiko Inami","en":""}],"title":{"ja":"MEME — Eye Wear Computing to Explore Human Behavior"},"journal":{"ja":"Proceedings of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing Adjunct Publication (UbiComp2015)"},"pages":"361-363","year":2015,"month":9,"location":"Osaka, Japan","review":"reviewed","presen_type":"poster","fields":[],"doi":"10.1145/2800835.2800900"},{"key":"Matsubara2015_ISWC","id":"1472","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Mizuki Matsubara","en":""},{"ja":"Joachim Folz","en":""},{"ja":"Takumi Toyama","en":""},{"ja":"Marcus Liwicki","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Extraction of Read Text for Automatic Video Annotation"},"journal":{"ja":"UbiComp/ISWC'15 Adjunct Proceedings of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of the 2015 ACM International Symposium on Wearable Computers"},"pages":"849-856","year":2015,"month":9,"location":"Osaka, Japan","review":"reviewed","presen_type":"oral","fields":["document-image-analysis"],"doi":"10.1145/2800835.2804333","abstract":{"ja":"This paper presents an automatic video annotation method which utilizes the user’s reading behaviour. Using a wearable eye tracker, we identify the video frames where the user reads a text document and extract the sentences that have been read by him or her. The extracted sentences are used to annotate video segments which are taken from the user’s egocentric perspective. An advantage of the proposed method is that we do not require training data, which is often used by a video annotation method. We examined the accuracy of the proposed annotation method with a pilot study where the experiment participants drew an illustration reading a tutorial. The method achieved 64:5% recall and 30:8% precision."}},{"key":"Kunze2015_Siggrapha","id":"1695","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Kai Kunze","en":"Kai Kunze"},{"ja":"Kazutaka Inoue","en":"Kazutaka Inoue"},{"ja":"Katsutoshi Masai","en":"Katsutoshi Masai"},{"ja":"Yuji Uema","en":"Yuji Uema"},{"ja":"Sean Shao-An Tsai","en":"Sean Shao-An Tsai"},{"ja":"Shoya Ishimaru","en":"Shoya Ishimaru"},{"ja":"Kastuma Tanaka","en":"Kastuma Tanaka"},{"ja":"Koichi Kise","en":"Koichi Kise"},{"ja":"Masahiko Inami","en":"Masahiko Inami"}],"title":{"ja":"Proceedings of ACM SIGGRAPH 2015 Emerging Technologies","en":"Proceedings of ACM SIGGRAPH 2015 Emerging Technologies"},"num_pages":"1","year":2015,"month":8,"review":"not_reviewed","fields":["sensing"]},{"key":"Karatzas2015","id":"1469","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Dimosthenis Karatzas","en":""},{"ja":"Lluis Gomez-Bigorda","en":""},{"ja":"Anguelos Nicolaou","en":""},{"ja":"Suman Ghosh","en":""},{"ja":"Andrew Bagdanov","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Jiri Matas","en":""},{"ja":"Lukas Neumann","en":""},{"ja":"Vijay Ramaseshan Chandrasekhar","en":""},{"ja":"Shijian Lu","en":""},{"ja":"Faisal Shafait","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Ernest Valveny","en":""}],"title":{"ja":"ICDAR 2015 Competition on Robust Reading"},"journal":{"ja":"Proc. 13th International Conference on Document Analysis and Recognition (ICDAR 2015)"},"pages":"1156-1160","year":2015,"month":8,"location":"Nancy, France","review":"reviewed","presen_type":"oral_and_poster","fields":["document-image-analysis","pattern-recognition"],"doi":"10.1109/ICDAR.2015.7333942","abstract":{"ja":"Results of the ICDAR 2015 Robust Reading Competition\nare presented. A new Challenge 4 on Incidental Scene\nText has been added to the Challenges on Born-Digital Images,\nFocused Scene Images and Video Text. Challenge 4 is run\non a newly acquired dataset of 1,670 images evaluating Text\nLocalisation, Word Recognition and End-to-End pipelines. In\naddition, the dataset for Challenge 3 on Video Text has been\nsubstantially updated with more video sequences and more\naccurate ground truth data. Finally, tasks assessing End-to-End\nsystem performance have been introduced to all Challenges. The\ncompetition took place in the first quarter of 2015, and received a\ntotal of 44 submissions. Only the tasks newly introduced in 2015\nare reported on. The datasets, the ground truth specification and\nthe evaluation protocols are presented together with the results\nand a brief summary of the participating methods."}},{"key":"Augereau2015_ICDAR","id":"1485","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Olivier Augereau","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Kensuke Hoshika","en":""}],"title":{"ja":"A proposal of a document image reading-life log based on document image retrieval and eyetracking"},"journal":{"ja":"Proc. of the 13th International Conference on Document Analysis and Recognition (ICDAR2015)"},"pages":"246 - 250","year":2015,"month":8,"review":"reviewed","fields":["reading-life-log"],"doi":"10.1109/ICDAR.2015.7333761","abstract":{"ja":"Instead of analyzing directly the document images, analyzing the document reading can offer new perspectives for extracting information about both the reader and the document. Analyzing how people read texts can help to understand the cognitive process of the reading and might lead to new approaches and new solutions for pattern recognition and document image analysis. It can also lead to create smart documents that can measure reading information, provide feedback and adapt themselves depending on the behavior of the readers. As a step towards document reading analysis, the authors propose in this paper a solution for extracting the reading information and creating a “reading-life log”. This reading-life log contains basic features that can be used for many different kinds of applications. A tag cloud evolving according to the reading is presented as a first application of the reading-life log.","en":"Instead of analyzing directly the document images, analyzing the document reading can offer new perspectives for extracting information about both the reader and the document. Analyzing how people read texts can help to understand the cognitive process of the reading and might lead to new approaches and new solutions for pattern recognition and document image analysis. It can also lead to create smart documents that can measure reading information, provide feedback and adapt themselves depending on the behavior of the readers. As a step towards document reading analysis, the authors propose in this paper a solution for extracting the reading information and creating a “reading-life log”. This reading-life log contains basic features that can be used for many different kinds of applications. A tag cloud evolving according to the reading is presented as a first application of the reading-life log."}},{"key":"Rigaud2015","id":"1482","type":"intl_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"","en":"Christophe Rigaud"},{"ja":"","en":"Nam Le Thanh"},{"ja":"","en":"J.-C. Burie"},{"ja":"","en":"J.-M. Ogier"},{"ja":"","en":"Motoi Iwata"},{"ja":"","en":"Eiki Imazu"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Speech balloon and speaker association for comics and manga understanding"},"journal":{"en":"Proc. of the 13th International Conference on Document Analysis and Recognition (ICDAR2015)"},"pages":"351-355","year":2015,"month":8,"review":"reviewed","fields":["document-image-analysis"]},{"key":"Kunze2015_SIGGRAPHb","id":"1696","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Kai Kunze","en":"Kai Kunze"},{"ja":"Kastutaka Inoue","en":"Kastutaka Inoue"},{"ja":"Katsutoshi Masai","en":"Katsutoshi Masai"},{"ja":"Yuji Uema","en":"Yuji Uema"},{"ja":"Sean Shao-An Tsai","en":"Sean Shao-An Tsai"},{"ja":"Shoya Ishimaru","en":"Shoya Ishimaru"},{"ja":"Katsuma Tanaka","en":"Katsuma Tanaka"},{"ja":"Koichi Kise","en":"Koichi Kise"},{"ja":"Masahiko Inami","en":"Masahiko Inami"}],"title":{"ja":"MEME: smart glasses to promote healthy habits for knowledge workers","en":"MEME: smart glasses to promote healthy habits for knowledge workers"},"journal":{"ja":"Proceedings of ACM SIGGRAPH 2015 Emerging Technologies","en":"Proceedings of ACM SIGGRAPH 2015 Emerging Technologies"},"pages":"1","year":2015,"month":8,"review":"reviewed","fields":["sensing"],"doi":"10.1145/2782782.2792491"},{"key":"Yoshimura2015","id":"1481","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Kazuyo Yoshimura","en":""},{"ja":"Kai Kunze","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"The Eye as the Window of the Language Ability: Estimation of English Skills by Analyzing Eye Movement While Reading Documents"},"journal":{"ja":"Proceedings of the 13th International Conference on Document Analysis and Recognition (ICDAR2015)"},"pages":"251-255","year":2015,"month":8,"review":"reviewed","fields":["reading-life-log"]},{"key":"Tanaka2015","id":"1699","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Katsuma Tanaka","en":""},{"ja":"Shoya Ishimaru","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Kai Kunze","en":""},{"ja":"Masahiko Inami","en":""}],"title":{"ja":"Nekoze!: monitoring and detecting head posture while working with laptop and mobile phone"},"journal":{"ja":"Proceedings of the 9th International Conference on Pervasive Computing Technologies for Healthcare"},"pages":"237-240","year":2015,"month":5,"fields":["sensing"]},{"key":"Ishimaru2015_ACM","id":"1459","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Shoya Ishimaru","en":"Shoya Ishimaru"},{"ja":"Kai Kunze","en":"Kai Kunze"},{"ja":"Katsuma Tanaka","en":"Katsuma Tanaka"},{"ja":"Yuji Uema","en":"Yuji Uema"},{"ja":"Koichi Kise","en":"Koichi Kise"},{"ja":"Masahiko Inami","en":"Masahiko Inami"}],"title":{"ja":"Smart Eyewear for Interaction and Activity Recognition","en":"Smart Eyewear for Interaction and Activity Recognition"},"journal":{"ja":"Proceedings of the 33rd Annual ACM Conference Extended Abstracts on Human Factors in Computing Systems","en":"Proceedings of the 33rd Annual ACM Conference Extended Abstracts on Human Factors in Computing Systems"},"pages":"307-310","year":2015,"month":4,"location":"Seoul, Republic of Korea","review":"reviewed","presen_type":"poster","fields":["pattern-recognition"],"doi":"10.1145/2702613.2725449","abstract":{"ja":"Smart glasses and, in general, eyewear are a fairly novel device class with a lot of possibilities for user interaction design and unobtrusive activity tracking. In this paper we show applications using an early prototype of J!NS MEME, smart glasses with integrated electrodes to detect eye movements (Electrooculography, EOG) and motion sensors (accelerometer and gyroscope) to monitor head motions. We present several demonstrations: We show a simple eye movement visualization, detecting left/right eye motion and blink. Additionally, users can play a game, ’Blinky Bird’. They need to help a bird avoid obstacles using eye movements. We implemented online detection of reading and talking behavior using a combination of blink, eye movement and head motion. We can give people a long term view of their reading, talking, and also walking activity over the day.","en":"Smart glasses and, in general, eyewear are a fairly novel device class with a lot of possibilities for user interaction design and unobtrusive activity tracking. In this paper we show applications using an early prototype of J!NS MEME, smart glasses with integrated electrodes to detect eye movements (Electrooculography, EOG) and motion sensors (accelerometer and gyroscope) to monitor head motions. We present several demonstrations: We show a simple eye movement visualization, detecting left/right eye motion and blink. Additionally, users can play a game, ’Blinky Bird’. They need to help a bird avoid obstacles using eye movements. We implemented online detection of reading and talking behavior using a combination of blink, eye movement and head motion. We can give people a long term view of their reading, talking, and also walking activity over the day."}},{"key":"Okoso2015_ACM","id":"1489","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Ayano Okoso","en":""},{"ja":"Joachim Folz","en":""},{"ja":"Takumi Toyama","en":""},{"ja":"Marcus Liwicki","en":""},{"ja":"Kai Kunze","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Towards Extraction of Subjective Reading Incomprehension: Analysis of Eye Gaze Features"},"journal":{"en":"Proceedings of the 33rd Annual ACM Conference Extended Abstracts on Human Factors in Computing Systems"},"year":2015,"month":4,"review":"reviewed","fields":[]},{"key":"Kunze2015_TAH_Augmentedb","id":"1694","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Kai Kunze","en":"Kai Kunze"},{"ja":"Susana Sanchez","en":"Susana Sanchez"},{"ja":"Tilman Dingler","en":"Tilman Dingler"},{"ja":"Olivier Augereau","en":"Olivier Augereau"},{"ja":"Koihchi Kise","en":"Koihchi Kise"},{"ja":"Masahiko Inami","en":"Masahiko Inami"},{"ja":"Terada Tsutomu","en":"Terada Tsutomu"}],"title":{"ja":"The Augmented Narrative ? Toward EstimatingReader Engagement","en":"The Augmented Narrative ? Toward EstimatingReader Engagement"},"journal":{"ja":"Proceedings of 6th Augmented Human International Conference","en":"Proceedings of 6th Augmented Human International Conference"},"pages":"163-164","year":2015,"month":3,"review":"not_reviewed","fields":["reading-life-log"]},{"key":"Kunze2015_TAH_Augmenteda","id":"1451","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"","en":"Kai Kunze"},{"ja":"","en":"Susana Sanchez"},{"ja":"","en":"Tilman Dingler"},{"ja":"","en":"Olivier Augereau"},{"ja":"","en":"Koichi Kise"},{"ja":"","en":"Masahiko Inami"},{"ja":"","en":"Tsutomu Terada"}],"title":{"en":"The Augmented Narrative — Toward Estimating Reader Engagement"},"journal":{"en":"Proceedings of the 6th Augmented Human International Conference"},"pages":"163-164","year":2015,"month":3,"location":"Singapore","review":"reviewed","presen_type":"oral","fields":[]},{"key":"Kise2015_SIG","id":"1525","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"クンツェ カイ","en":"Kai Kunze"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"稲見 昌彦","en":"Masahiko Inami"}],"title":{"ja":"Reading-Life Logへの挑戦"},"journal":{"ja":"情報処理学会研究報告"},"volume":"2015-DC-100","number":"4","pages":"1-8","year":2015,"month":12,"review":"not_reviewed","fields":["reading-life-log"]},{"key":"Augereau2015_MIRU","id":"1466","type":"domestic_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Olivier Augereau","en":""},{"ja":"Kai Kunze","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Estimating the number of read words with the Wordometer"},"journal":{"ja":"MIRU 2015 Demo"},"year":2015,"month":8,"location":"大阪府吹田市","review":"not_reviewed","presen_type":"poster","fields":["document-image-analysis","pattern-recognition"]},{"key":"Inadomi2015","id":"1507","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"稲富 宏之","en":"Hiroyuki Inadomi"},{"ja":"岩田 基","en":"Motoi Iwata"},{"ja":"田中 宏明","en":"Hiroaki Tanaka"},{"ja":"奥谷 研","en":"Ken Okutani"},{"ja":"伊藤 斉子","en":"Saiko Ito"}],"title":{"ja":"健康増進活動に参加する地域在住の高齢者にUBOMを用いた経験"},"journal":{"ja":"UBOM（簡易客観的精神指標検査）技術講習会・2015"},"year":2015,"month":8,"review":"not_reviewed","fields":[]},{"key":"Fujiyoshi2015_PRMU","id":"1462","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"藤好 宏樹","en":""},{"ja":"吉村 和代","en":""},{"ja":"Kai Kunze","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"英文問題解答時の視点情報を用いた英語能力推定法"},"book_title":"電子情報通信学会技術研究報告","volume":"115","number":"24","presen_id":"PRMU2015-10","pages":"pp.49-54","year":2015,"month":5,"location":"三重県津市","review":"not_reviewed","presen_type":"oral","fields":[],"abstract":{"ja":"本稿では，アイトラッカで取得した英文問題解答時の視点情報を用いて，英語習熟度を推定する手法を提\n案する．一般に，英語習熟度を測るテストとしてTOEIC が挙げられる．しかし，TOEIC には結果が出るまでに時間\nを要する，問題数が多く受験者に負担が大きいという課題がある．そこで，TOEIC よりも少ない問題数で手軽に英\n語習熟度を測る事ができれば有益であると考えられる．このような推定を可能にするための方法として，テストの問\n題正答数の他に，問題解答時の視点情報を利用することが考えられる．本手法では眼球運動の一種である\ffixationとsaccadeに関する特徴を利用する．英文問題10 題を用いて被験者9 人の英語習熟度を推定したところ，TOEIC スコ\nアを平均絶対誤差36.3 点で推定することができた。"}},{"key":"Hoshika2015","id":"1498","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"星加 健介","en":""},{"ja":"藤好 宏樹","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"視点情報に基づくリーディングライフログの取得とその視覚化"},"book_title":"電子情報通信学会技術研究報告","volume":"115","number":"24","presen_id":"PRMU2015-9","pages":"43-48","year":2015,"month":5,"location":"三重県津市","review":"not_reviewed","presen_type":"oral","fields":[],"abstract":{"ja":"本稿では，視点情報に基づくユーザ特性を付与したリーディングライフログの視覚化を提案する．リーディ ングライフログとは，人の読む行動を記録することであり，様々な研究が行われている．本稿では，リーディングラ イフログの中でも，ユーザがいつ，どの単語を読んだかという詳細なログの取得を目指す．このようなログを取得で きれば，ユーザの読んだ単語を検索しやすくなる．ユーザの読んだ単語を特定するためには，アイトラッカを用いて， 視点の位置を推定する．推定した視点の位置とユーザが本当に見ている視点の位置には誤差を含むので，重み付けに よって，誤差を考慮した推定の精度を取得する．時刻毎に記録した膨大なログを見ただけで，内容を理解することは 難しい．そこで，Wordle を用いて視覚化し，ユーザ自身の単語のログを一目で理解できるように工夫する．文書の単 語出現回数に基づく Wordle と作成した視点情報に基づく Wordle を比較した．比較した結果，表示される単語の大き さの違いを確認できた．また，被験者間の Wordle は異なることを確認し，視点の動きを観察した．"}},{"key":"Okoso2015_PRMU","id":"1488","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"大社 綾乃","en":""},{"ja":"Kai Kunze","en":""},{"ja":"Olivier Augereau","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"学習補助のための視点情報に基づく文書アノテーション"},"book_title":"電子情報通信学会技術研究報告","volume":"115","number":"24","presen_id":"PRMU2015-30","pages":"161-166","year":2015,"month":5,"review":"not_reviewed","fields":[],"abstract":{"ja":"本稿では，アイトラッカから取得した視点情報を用いて，文書を読んだ際に難しいと感じた部分を推定し，その情報を視覚化するシステムを提案する．文書を読むという行為から，学習者の理解できていない部分を推定することが出来れば，次の２点で大変有用であると考えられる．まず，学習者は普段の学習において，難しいと感じた部分の全てに印をつける手間を省くことが出来る．そして，指導者はその情報をもとに，学習者に合った指導が可能になる．一般的に人は文書を読む際，難しいと感じると読むスピードが落ちたり，頻繁に読み返す，同じ場所を注視し続けるといった傾向がある．本システムでは，このような特性から読者が難しいと感じた部分を推定し，その結果をもとに，段落ごとおよび単語ごとに文書画像上にアノテーションを付与する．被験者８人，１０文書を用いて，読書中の視点情報から，難しいと感じた部分を推定し，アノテーションを作成したところ，（１）読む速さや注視回数は被験者間の理解度の違いの識別に有効であり，（2）注視時間が難しいと感じた部分の識別に有効であることが分かった．"}},{"key":"Miyagawa2015_CVIM","id":"1461","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"宮川 拓也","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"顔認識における特定物体認識と一般物体認識の比較"},"journal":{"ja":"情報処理学会研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2015-CVIM-197","pages":"1-7","year":2015,"month":5,"publisher":"一般社団法人情報処理学会","review":"not_reviewed","fields":["object-recognition"],"note":{"ja":"ここに掲載した著作物の利用に関する注意 本著作物の著作権は情報処理学会に帰属します。本著作物は著作権者である情報処理学会の許可のもとに掲載するものです。ご利用に当たっては「著作権法」ならびに「情報処理学会倫理綱領」に従うことをお願いいたします。","en":"Notice for the use of this material The copyright of this material is retained by the Information Processing Society of Japan (IPSJ). This material is published on this web site with the agreement of the author (s) and the IPSJ. Please be complied with Copyright Law of Japan and the Code of Ethics of the IPSJ if any users wish to reproduce, make derivative work, distribute or make available to the public any part or whole thereof.\nAll Rights Reserved, Copyright (C) Information Processing Society of Japan.\nComments are welcome. Mail to address editj＠ipsj.or.jp, please."},"abstract":{"ja":"物体認識手法は扱う問題によって特定物体認識と一般物体認識の二種類が使い分けられる．しかし顔認識は両方のアプローチで高精度な顔認識に成功しており，どちらのアプローチが適しているのかは明らかにされていない．どちらかのアプローチの優位性や，それぞれのアプローチが特に有効なタスクなどを示すことができれば，今後の顔認識技術の発展につながると考える．本論文では既存手法を利用した比較実験を通して，顔認識における特定物体認識と一般物体認識それぞれの有用性について考察する．また副題として，顔認識における各要素の重要度を示すため，一般物体認識による顔認識においてどのようなvisual wordsが有効であるのかを調べる．実験の結果，認識率の低かった表情変化の大きいクエリについて他のクエリよりも大きな差が生まれ，一般物体認識の方が高い認識率を示した．visual wordsに関する検証実験では，顔認識において位置情報が必要不可欠な要素であるということを示した．"}},{"key":"Mizuno2015_CVIM","id":"1460","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"水野 智也","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"Bucket Distance HashingとMetric Learningを組み合わせた表情変化に頑健かつ高速な顔認識"},"journal":{"ja":"情報処理学会研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2015-CVIM-197","pages":"1-5","year":2015,"month":5,"review":"not_reviewed","fields":["pattern-recognition"],"note":{"ja":"ここに掲載した著作物の利用に関する注意 本著作物の著作権は情報処理学会に帰属します。本著作物は著作権者である情報処理学会の許可のもとに掲載するものです。ご利用に当たっては「著作権法」ならびに「情報処理学会倫理綱領」に従うことをお願いいたします。","en":"Notice for the use of this material The copyright of this material is retained by the Information Processing Society of Japan (IPSJ). This material is published on this web site with the agreement of the author (s) and the IPSJ. Please be complied with Copyright Law of Japan and the Code of Ethics of the IPSJ if any users wish to reproduce, make derivative work, distribute or make available to the public any part or whole thereof.\nAll Rights Reserved, Copyright (C) Information Processing Society of Japan.\nComments are welcome. Mail to address editj＠ipsj.or.jp, please."},"abstract":{"ja":"大規模なデータベース(DB) に対して高速な顔認識手法の一つに内海らの手法がある．内海らの手法はクエリの特徴量とのユークリッド距離が近いDBの特徴量を高速に探索する近似最近傍探索手法を用いることで高速な顔認識を実現した．しかし，内海らの手法はクエリの特徴量と同一の特徴量を探索する手法であるため，表情変化に弱いという問題がある．表情変化に頑健な顔認識手法にCaoらの手法がある．Caoらの手法はMetric Learningを用いることで，表情変化に頑健な顔認識を実現した．しかし，Caoらの手法には処理速度が遅いという問題がある．本稿では，表情変化に頑健なCaoらの手法に内海らの手法で使用されている高速な近似最近傍探索手法を導入する．内海らの手法とCaoらの手法では評価関数が異なるため，近似最近傍探索を直接Caoらの手法に適応できない．そこで，Caoらの類似度評価関数をユークリッド距離で表現することで，高速化を実現する．100枚の顔画像DBでの顔認識実験を行った結果，内海らの手法と比べて認識速度を保ったまま，認識率を10%上昇させることに成功した．"}},{"key":"Eiki2015_PRMU","id":"1446","type":"domestic_conf","language":"JA","show":"show_except_abstract","authors":[{"ja":"今津英貴","en":""},{"ja":"岩田基","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"漫画キャラクターへの固有ラベル付けのための学習方法の検討"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"114","number":"521","presen_id":"PRMU2014-185","pages":"159-164","year":2015,"month":3,"review":"not_reviewed","fields":["pattern-recognition","information-retrieval"]},{"key":"Kitano2015","id":"1437","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"北野 豊","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"解像度低下と照明変動に頑健な大規模顔画像認識"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"114","number":"521","presen_id":"2014-192","pages":"199-204","year":2015,"month":3,"review":"not_reviewed","fields":[],"note":{"ja":"発表原稿（prmu_kitano.pdf）の表１，表２には誤りがあります．訂正したものがprmu_kitano_re.pdfです．"},"abstract":{"ja":"数百万枚の顔画像を含む大規模データベースに対して，高速・高精度な顔認識が可能な手法として、坂野らの手法がある．この手法では，局所特徴の検出器を使用して，画像に応じて特徴抽出箇所を決定する．検出器は低解像度顔画像や照明変化に弱いため，坂野らの手法もこれらの画像変化を受けた顔画像の認識に弱いという欠点がある．この問題を解決する方法として，局所特徴の検出方法として\nあらかじめ定められた位置から特徴点を密にとるDense Samplingを用いることが考えられる．しかし，これではメモリ使用量が大幅に増加し，また表情変化に弱くなる．そこで本稿では，Dense Samplingを導入することで生じるメモリ使用量と表情変化の問題を同時に解決する方法を提案する．これにより，低解像度顔画像や照明変化を受けた顔画像の認識性能を向上しつつ，省メモリで表情変化に頑\n健な認識を実現する．"}},{"key":"Tsuyama2015_PRMU","id":"1436","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"津山 裕加","en":"Yuka Tsuyama"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"画像処理による単純な文字の特徴の増加手法の提案","en":"Obtaining more feature points from simple characters"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"114","number":"521","presen_id":"PRMU2014-163","pages":"23-28","year":2015,"month":3,"location":"神奈川県横浜市","review":"not_reviewed","presen_type":"oral","fields":[],"note":{"ja":"発表原稿（paper.pdf）は表3に誤りがあります． それを修正したものがpaper_revised.pdfになります．"},"abstract":{"ja":"近年，情景中の文字を認識する技術についての研究が盛んに行われている．\n既存の情景内文字認識手法の一つである松田らの手法では，画像から得られる局所特徴を用いることで，\n複雑背景・レイアウト上の文字でも高精度で認識することができる．\nしかし，数字やアルファベットといった単純な文字は，局所特徴が十分に得られないことから認識率が低い．\nそこで本稿では，認識対象の画像に画像処理を施し，文字の特徴を残したままぼかすことによって文字領域から有効な特徴をより多く抽出し，単純な文字の認識を可能にする手法を提案する．\n実験の結果，ぼかすことで新たに認識できる文字があることが分かった．"}},{"key":"Kazuyo2015_PRMU","id":"1435","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"吉村 和代","en":"Yoshimura Kazuyo"},{"ja":"Kai Kunze","en":"Kai Kunze"},{"ja":"黄瀬 浩一","en":"Kise Koichi"}],"title":{"ja":"読書時の眼球運動を利用した英語習熟度推定法"},"book_title":"電子情報通信学会技術研究報告","volume":"114","number":"454","presen_id":"PRMU2014-123","pages":"63-68","year":2015,"month":2,"location":"宮城県仙台市","review":"not_reviewed","presen_type":"oral_and_poster","fields":[]},{"key":"Konishi2015_PRMU","id":"1432","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"小西 將貴","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"特徴空間の次元数と部分空間数にスケーラブルな近似最近傍部分空間探索"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"114","number":"454","presen_id":"PRMU2014-124","pages":"69-74","year":2015,"month":2,"location":"宮城県仙台市","review":"not_reviewed","presen_type":"oral_and_poster","fields":["pattern-recognition","nearest-neighbor-search"],"awards":{"ja":["平成26（2014）年度 PRMU研究奨励賞受賞"],"en":["2014 PRMU Award Winner"]},"note":{"ja":"paper.pdfは図4(c)に誤りがあります．\nそれを修正したものがpaper_revised.pdfになります．"},"abstract":{"ja":"パターンを部分空間で表現する方法がコンピュータビジョンやパターン認識の分野で広く利用されている．\n扱う問題の大規模化に伴い，最も近い部分空間を探索する最近傍部分空間探索のスケーラビリティの向上が求められている．\nその障害になるのは，部分空間同士の距離はグラスマン多様体上で定義される距離で測られるため，ユークリッド空間上の距離でよく用いられる近似最近傍探索手法が直接適用できないことである．\n近似を導入した部分空間の効率的な探索を目指す手法は提案されているが，大規模なデータセットや高次元のデータに有効ではない．\n本稿では，グラスマン多様体上の距離の計算を複数のユークリッド距離の計算に分解することで，部分空間同士の類似度の効率的な近似計算を実現する．実験の結果，部分空間の探索精度を3% 落とすことで従来手法と比べて，約800 倍の高速化が実現できた．"}},{"key":"Anpo2015","id":"2063","type":"book","language":"JA","show":"show","authors":[{"ja":"安保 正一","en":""},{"ja":"福田 弘和","en":""},{"ja":"和田 光生","en":""},{"ja":"西浦 芳史","en":""},{"ja":"古川 一","en":""},{"ja":"小倉 東一","en":""},{"ja":"福田 直也","en":""},{"ja":"東 孝信","en":""},{"ja":"秋間 和広","en":""},{"ja":"明神 千穂","en":""},{"ja":"西野 慎一","en":""},{"ja":"亀岡 孝治","en":""},{"ja":"伊能 利郎","en":""},{"ja":"大原 均","en":""},{"ja":"中村 謙治","en":""},{"ja":"西田 良和","en":""},{"ja":"横山 良平","en":""},{"ja":"青木 考","en":""},{"ja":"羽藤 堅治","en":""},{"ja":"杉村 延広","en":""},{"ja":"岩村 幸治","en":""},{"ja":"菅野 正嗣","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"木村 一貫","en":""},{"ja":"守行 正悟","en":""},{"ja":"吉田 篤正","en":""},{"ja":"木下 進一","en":""},{"ja":"兼子まや","en":""},{"ja":"奥田 邦晴","en":""},{"ja":"片岡 正教","en":""},{"ja":"岡原 聡","en":""},{"ja":"宮垣 慶子","en":""},{"ja":"丸尾 達","en":""},{"ja":"浦出 俊和","en":""},{"ja":"林 意雀","en":""},{"ja":"王 達","en":""}],"title":{},"book_title":"植物工場の生産性向上,コスト削減技術とビジネス構築-今,注目されている植物工場とは,その全容-","year":2015,"month":11,"publisher":"シーエムシー出版","review":"not_reviewed","fields":["plant-image-processing"],"url":"https://www.cmcbooks.co.jp/products/detail.php?product_id=7607"},{"key":"Matsuda2015","id":"1776","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"松田 崇宏","en":""}],"title":{"ja":"局所特徴ベース情景内文字認識手法の性能向上","en":"Performance Improvement in Local Feature Based Scene Character Recognition"},"year":2015,"month":3,"fields":[]},{"key":"Miyazawa2015","id":"1438","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"宮沢嘉希","en":""}],"title":{"ja":"早期棄却と早期認識を用いた文書画像検索法"},"year":2015,"month":3,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval","object-recognition"]},{"key":"Riki2015","id":"1433","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"工藤 力","en":"Kudo Riki"}],"title":{"ja":"一般的な筆記に適合したカメラペンシステム","en":"Camera—pen System Applicable to Natural Handwriting"},"year":2015,"month":3,"review":"not_reviewed","fields":[]},{"key":"Kise2015_J","id":"1452","type":"survey","language":"JA","show":"show_except_datafile","authors":[{"ja":"黄瀬浩一","en":"Koichi KISE"},{"ja":"大町真一郎","en":"Shinichiro OMACHI"},{"ja":"内田誠一","en":"Seiichi UCHIDA"},{"ja":"岩村雅一","en":"Masakazu IWAMURA"}],"title":{"ja":"文字・文書メディアの新しい利用基盤技術の開発への取組み","en":"A Trial for development of fundamental technologies for new usage of character and document media"},"journal":{"ja":"電子情報通信学会誌","en":"The Journal of the Institute of Electronics, Information and Communication Engineers"},"volume":"98","number":"4","pages":"311-327","year":2015,"month":4,"review":"not_reviewed","fields":["document-image-analysis","pattern-recognition"],"url":"http://www.journal.ieice.org/summary.php?id=k98_4_311&year=2015&lang=J","abstract":{"ja":"“You are what you read.”と言われるように，読むことは，我々の知識や人格の形成に中心的な役割を果たしている．実際，知識を追加・更新するため，我々は毎日読むことに膨大な時間を費やしている．ところが，このような努力は記録も再利用もされておらず，かけた時間に見合った価値を引き出せているとは言い難い．本稿では，読むという行動を記録し，再利用するためのリーディングライフログ（Reading-Life Log）技術について解説する．具体的には，人の読む行動と読まれる対象を相互解析することによって，読んだ文字・文書を通して人を知り，また人の読み方を通して文字・文書を知ることを実現する．"}},{"key":"Kise2015_ICDAR","id":"1508","type":"other","language":"EN","show":"show","authors":[{"ja":"","en":"Koichi Kise"}],"title":{"en":"Document Analysis Meets  Activity Recognition:  A New Paradigm of Analyzing Documents in Combination With Users’ Reading Behavior"},"journal":{"en":"13th International Conference on Document Analysis and Recognition (ICDAR2015)"},"year":2015,"month":8,"review":"not_reviewed","fields":["document-image-analysis"],"note":{"en":"Invited Keynote Talk"}},{"key":"Utsumi2015","id":"1926","type":"other","language":"JA","show":"show","authors":[{"ja":"内海 ゆづ子","en":""}],"title":{"ja":"輝け！リケジョ(理系女子)(第17回) 研究者になるまでとこれから"},"journal":{"ja":"映像情報メディア学会誌"},"volume":"69","number":"1","pages":"87-89","year":2015,"month":1,"review":"not_reviewed","fields":[],"doi":"10.3169/itej.69.87"},{"key":"Iwata2014_J","id":"1426","type":"journal_paper","language":"JA","show":"show_except_datafile","authors":[{"ja":"岩田 基","en":"Motoi Iwata"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"透かしの痕跡を利用して再撮影動画像から透かし入り動画像の領域を推定する電子透かし法の実装実験","en":"Digital Watermarking Method Estimating Watermarked Video Region from Recaptured Video using Trace of Watermark"},"journal":{"ja":"日本画像学会誌","en":"The Imaging Society of Japan"},"volume":"53","number":"5","pages":"436-447","year":2014,"month":10,"review":"reviewed","fields":[]},{"key":"Utsumi2014_IEICE","id":"1410","type":"journal_paper","language":"JA","show":"show_except_datafile","authors":[{"ja":"内海 ゆづ子","en":"Yuzuko Utsumi"},{"ja":"坂野 悠司","en":"Yuji Sakano"},{"ja":"前川 敬介","en":"Keisuke Maekawa"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"局所特徴量と投票処理を用いた大規模データベースに対する高速顔認識","en":"Fast Face Recognition Based on Local Features and Voting Scheme"},"journal":{"ja":"電子情報通信学会論文誌D","en":"Trans. IEICE"},"volume":"J97-D","number":"8","pages":"1263-1272","year":2014,"month":8,"review":"reviewed","fields":["pattern-recognition"],"url":"http://search.ieice.org/bin/summary.php?id=j97-d_8_1263","note":{"ja":"論文のPDFファイルを<a href=\"http://www.ieice.org/jpn/trans_online/\">電子情報通信学会 Transactions online</a>からダウンロードすることができます．","en":"PDF file is available at <a href=\"http://www.ieice.org/jpn/trans_online/\">IEICE Transactions online (in Japanese)</a>."},"abstract":{"ja":"犯罪捜査や写真の自動整理などでは，大量の顔画像から特定の人物を探し出すことが求められる．\nそこで問題となるのは，認識したい顔画像を顔認識システムに登録された大量の顔画像と照合するのにかかる時\n間と認識精度である．既存手法の多くは大量の顔画像の処理を想定せずに設計されているため，登録画像が少数\nの場合は高い認識率を達成できるが，登録画像が増加すると実用的な時間で計算が終わらないという問題がある．\n高速性と認識精度を兼ね備えた顔認識システムを実現する有効な対策として，高速な手法で正解候補を少数に絞\nりこみ，絞り込まれた候補に対して既存の手法を適用するという方法がある． 本論文では，顔認識の精度を保ち\nつつ高速化を実現することを目的とし，大規模データベースから高速に候補を絞り込む顔認識手法を提案する．\n提案手法は局所特徴量，近似最近傍探索，投票を組み合わせることによって，これを実現する．","en":"Personal authentication for criminal investigations and immigration control requires finding a particular person from massive face images. The most important elements of the personal authentication are computational time and accuracy of recognition when massive images are registered in a face\nrecognition system, namely scalability against the size of registered information. Many conventional face\nrecognition methods can recognize face images accurately. However, if the number of registered images in-\ncreases, computational time for recognition increases in proportion to the number of registered images and\nit is impossible to execute face recognition in real-time. This is because the conventional methods did not\ntake the scalability into account. If we narrow the candidates of recognition down from registered images\nand apply the conventional methods to the narrowed candidates, we can realize a fast and high-accuracy\nface recognition system. In this paper, we propose a fast face recognition method which narrow down the candidates from massive face images by using local features, an approximate nearest neighbor method and a voting scheme. We made a 3 million face database and used the database as learning samples to evaluate the proposed method. The proposed method narrowed down recognition candidates to 1000 images with 257 ms and 96.9% accuracy."}},{"key":"Iwamura2014_CVA","id":"1404","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Masashi Imura","en":""},{"ja":"Shinsaku Hiura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Recognition of Defocused Patterns"},"journal":{"ja":"IPSJ Transactions on Computer Vision and Applicaitons (CVA)"},"volume":"6","pages":"48-52","year":2014,"month":7,"review":"reviewed","fields":["pattern-recognition","computational-photography"],"doi":"10.2197/ipsjtcva.6.48","abstract":{"ja":"The paper addresses the recognition problem of defocused patterns. Though recognition algorithms assume that the input images are focused and sharp, it does not always hold on actual camera-captured images. Thus, a recognition method that can recognize defocused patterns is required. In this paper, we propose a novel recognition framework for defocused patterns, relying on a single camera without a depth sensor. The framework is based on the coded aperture which can recover a less-degraded image from a defocused image if depth is available. However, in the problem setting of “a single camera without a depth sensor, ” estimating depth is ill-posed and an assumption is required to estimate the depth. To solve the problem, we introduce a new assumption suitable for pattern recognition; templates are known. It is based on the fact that in pattern recognition, all templates must be available in advance for training. The experiments confirmed that the proposed method is fast and robust to defocus and scaling, especially for heavily defocused patterns."}},{"key":"Chikano2014","id":"1301","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Megumi Chikano","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Shinichiro Omachi","en":""}],"title":{"ja":"Recovery and localization of handwritings by a camera-pen based on tracking and document image retrieval"},"journal":{"ja":"Pattern Recognition Letters"},"volume":"35","number":"1","pages":"214-224","year":2014,"month":1,"review":"reviewed","fields":["document-image-analysis","document-image-retrieval"],"doi":"10.1016/j.patrec.2012.10.003","abstract":{"ja":"We propose a camera-based method for digital recovery of handwritings on ordinary paper. Our method is characterized by the following two points: (1) it requires no special device such as special paper other than a camera-pen to recover handwritings, (2) if the handwriting is on a printed document, the method is capable of localizing it onto an electronic equivalent of the printed document. The above points are enabled by the following processing. The handwriting is recovered by the LK tracking to trace the move of the pen-tip. The recovered shape is localized onto the corresponding part of the electronic document with the help of document image retrieval called LLAH (locally likely arrangement hashing). A new framework for stably estimating the homography from a camera-captured image to the corresponding electronic document allows us to localize the recovered handwritings accurately. We experimentally evaluate the accuracy, processing time and memory usage of the proposed method using 30 handwritings. From the comparison to other methods that implement alternative ways for realizing the same functionality, we have confirmed that the proposed method is superior to those other methods."}},{"key":"Liwicki2014","id":"1298","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Marcus Liwicki","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Akira Yoshida","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"More than Ink — Realization of a Data-Embedding Pen"},"journal":{"ja":"Pattern Recognition Letters"},"volume":"35","number":"1","pages":"246-255","year":2014,"month":1,"review":"reviewed","fields":["document-image-analysis"],"doi":"10.1016/j.patrec.2012.09.001","abstract":{"ja":"In this paper we present a novel digital pen device, called data-embedding pen, for enhancing the value of handwriting on physical paper. This pen produces an additional ink-dot sequence along a written stroke during writing. This ink-dot sequence represents arbitrary information, such as writer’s name and writing date. Since the information is placed on the paper as an ink-dot sequence, it can be retrieved just by scanning or photographing the paper. In addition to the hardware of the data-embedding pen, this paper also proposes a coding scheme for reliable data-embedding and retrieval. In fact, the physical data-embedding on a paper will undergo various severe errors and therefore a robust coding scheme is necessary. Through experiments on data written by two writers, we show that we can embed 32 bits on short and simple or even on more complex patterns and finally retrieve them with a high reliability."}},{"key":"Shiga2014_Trans","id":"1418","type":"letter","language":"JA","show":"show_except_datafile","authors":[{"ja":"志賀 優毅","en":"Yuki SHIGA"},{"ja":"内海 ゆづ子","en":"Yuzuko UTSUMI"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"},{"ja":"カイ クンツェ","en":"Kai KUNZE"},{"ja":"黄瀬 浩一","en":"Koichi KISE"}],"title":{"ja":"読書活動の自動的記録のための文書画像の識別","en":"Document Type Classification for Automatic Recording of Read- ing Activities"},"journal":{"ja":"電子情報通信学会論文誌D"},"volume":"J97-D","number":"12","pages":"1733-1736","year":2014,"month":12,"review":"reviewed","fields":["document-image-analysis"],"url":"http://search.ieice.org/bin/summary.php?id=j97-d_12_1733","abstract":{"ja":"読書活動を把握することは，人の知識の質と量の把握につながるため，重要であると考えられている．とりわけ，多くの人の読書活動を長期間に渡って詳細に把握することができれば，これまでにない知見が得られる可能性がある．\nしかし，読書活動を記録する現在利用可能な方法は手動記録に限られ，現実的に掛けられる手間を考えると，詳細な読書活動の記録には適していない．\nそこで本論文では，読書活動を自動的に記録，分析するために，ユーザが読んでいる文書を推定する問題を考え，その意義と課題を議論する．"}},{"key":"Iwamura2014_ACM","id":"1429","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Nobuaki Matozaki","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Fast Instance Search Based on Approximate Bichromatic Reverse Nearest Neighbor Search"},"journal":{"ja":"Proc. of the ACM International Conference on Multimedia (MM '14)"},"pages":"1121-1124","year":2014,"month":11,"location":"Orlando, USA","review":"reviewed","presen_type":"poster","fields":["nearest-neighbor-search","object-recognition"],"doi":"10.1145/2647868.2654988","note":{"ja":"http://dl.acm.org/citation.cfm?id=2654988"}},{"key":"Shiga2014_PETMEI","id":"1420","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"","en":"Yuki Shiga"},{"ja":"","en":"Takumi Toyama"},{"ja":"","en":"Yuzuko Utsumi"},{"ja":"","en":"Andreas Dengel"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Daily Activity Recognition Combining Gaze Motion and Visual Features"},"journal":{"en":"PETMEI 2014: The 4th International Workshop on Pervasive Eye Tracking and Mobile Eye-based Interaction"},"book_title":"Proceedings of the 16th International Conference on Ubiquitous Computing","pages":"1103-1111","year":2014,"month":9,"location":"Seattle, USA","review":"reviewed","presen_type":"oral","fields":[],"doi":"10.1145/2638728.2641691"},{"key":"Ishimaru2014_UbiComp_Position","id":"1423","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Shoya Ishimaru","en":"Shoya Ishimaru"},{"ja":"Kai Kunze","en":"Kai Kunze"},{"ja":"Koichi Kise","en":"Koichi Kise"},{"ja":"Masahiko Inami","en":"Masahiko Inami"}],"title":{"ja":"Position Paper: Brain Teasers — Toward Wearable Computing that Engages Our Mind","en":"Position Paper: Brain Teasers — Toward Wearable Computing that Engages Our Mind"},"journal":{"ja":"Proceedings of the 2014 ACM Conference on pervasive and ubiquitous computing adjunct publication (UbiComp2014)","en":"Proceedings of the 2014 ACM Conference on pervasive and ubiquitous computing adjunct publication (UbiComp2014)"},"pages":"1405-1408","year":2014,"month":9,"location":"Seattle, USA","review":"reviewed","presen_type":"oral","fields":["pattern-recognition"],"doi":"10.1145/2638728.2641722","abstract":{"ja":"The emerging field of cognitive activity recognition – real life tracking of mental states – can give us new possibilities to enhance our minds.\nIn this paper, we outline the use of wearable computing to engage the user’s mind. We argue that the more personal technology becomes the more it should also adopt to the user’s long term goals improving mental fitness. We present a the concept of computing to engage our minds, discuss some enabling technologies as well as challenges and opportunities.","en":"The emerging field of cognitive activity recognition – real life tracking of mental states – can give us new possibilities to enhance our minds.\nIn this paper, we outline the use of wearable computing to engage the user’s mind. We argue that the more personal technology becomes the more it should also adopt to the user’s long term goals improving mental fitness. We present a the concept of computing to engage our minds, discuss some enabling technologies as well as challenges and opportunities."}},{"key":"Ishimaru2014_UbiComp_Shiny","id":"1422","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Shoya Ishimaru","en":"Shoya Ishimaru"},{"ja":"Jens Weppner","en":"Jens Weppner"},{"ja":"Andreas Poxrucker","en":"Andreas Poxrucker"},{"ja":"Paul Lukowicz","en":"Paul Lukowicz"},{"ja":"Kai Kunze","en":"Kai Kunze"},{"ja":"Koichi Kise","en":"Koichi Kise"}],"title":{"ja":"Shiny — An Activity Logging Platform for Google Glass","en":"Shiny — An Activity Logging Platform for Google Glass"},"journal":{"ja":"Proceedings of the 2014 ACM Conference on pervasive and ubiquitous computing adjunct publication (UbiComp2014)","en":"Proceedings of the 2014 ACM Conference on pervasive and ubiquitous computing adjunct publication (UbiComp2014)"},"pages":"283-286","year":2014,"month":9,"location":"Seattle, USA","review":"reviewed","presen_type":"oral","fields":["pattern-recognition"],"doi":"10.1145/2638728.2638798","abstract":{"ja":"We describe an activity logging platform for Google Glass based on our previous work. We introduce new multi-modal methods for quick non-disturbing interactions for activity logging control and real time ground truth labeling, consisting of swipe gesture, head gesture and laser pointer tagging methods. The methods are evaluated in user studies towards estimating their effectiveness.","en":"We describe an activity logging platform for Google Glass based on our previous work. We introduce new multi-modal methods for quick non-disturbing interactions for activity logging control and real time ground truth labeling, consisting of swipe gesture, head gesture and laser pointer tagging methods. The methods are evaluated in user studies towards estimating their effectiveness."}},{"key":"Ishimaru2014_UbiComp_Smarter","id":"1421","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Shoya Ishimaru","en":"Shoya Ishimaru"},{"ja":"Kai Kunze","en":"Kai Kunze"},{"ja":"Katsuma Tanaka","en":"Katsuma Tanaka"},{"ja":"Yuji Uema","en":"Yuji Uema"},{"ja":"Koichi Kise","en":"Koichi Kise"},{"ja":"Masahiko Inami","en":"Masahiko Inami"}],"title":{"ja":"Smarter Eyewear — Using Commercial EOG Glasses for Activity Recognition","en":"Smarter Eyewear — Using Commercial EOG Glasses for Activity Recognition"},"journal":{"ja":"Proceedings of the 2014 ACM Conference on pervasive and ubiquitous computing adjunct publication (UbiComp2014)","en":"Proceedings of the 2014 ACM Conference on pervasive and ubiquitous computing adjunct publication (UbiComp2014)"},"pages":"239-242","year":2014,"month":9,"location":"Seattle, USA","review":"reviewed","presen_type":"oral_and_poster","fields":["pattern-recognition"],"doi":"10.1145/2638728.2638795","abstract":{"ja":"Smart eyewear computing is a relatively new subcategory in ubiquitous computing research, which has enormous potential. In this paper we present a first evaluation of soon commercially available Electrooculography (EOG) glasses (J!NS MEME) for the use in activity recognition. We discuss the potential of EOG glasses and other smart eye-wear. Afterwards, we show a first signal level assessment of MEME, and present a classification task using the glasses. We are able to distinguish of 4 activities for 2 users ( typing, reading, eating and talking) using the sensor data (EOG and acceleration) from the glasses with an accuracy of 70 % for 6 sec. windows and up to 100 % for a 1 minute majority decision. The classification is done user-independent.","en":"Smart eyewear computing is a relatively new subcategory in ubiquitous computing research, which has enormous potential. In this paper we present a first evaluation of soon commercially available Electrooculography (EOG) glasses (J!NS MEME) for the use in activity recognition. We discuss the potential of EOG glasses and other smart eye-wear. Afterwards, we show a first signal level assessment of MEME, and present a classification task using the glasses. We are able to distinguish of 4 activities for 2 users ( typing, reading, eating and talking) using the sensor data (EOG and acceleration) from the glasses with an accuracy of 70 % for 6 sec. windows and up to 100 % for a 1 minute majority decision. The classification is done user-independent."}},{"key":"Okoso2014","id":"1448","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Ayano Okoso","en":""},{"ja":"Kai Kunze","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Implicit gaze based annotations to support second language learning"},"journal":{"ja":"Proceedings of the 2014 ACM Conference on pervasive and ubiquitous computing adjunct publication (UbiComp2014)"},"pages":"143-146","year":2014,"month":9,"location":"Seattle, USA","review":"reviewed","presen_type":"poster","fields":[],"doi":"10.1145/2638728.2638783","url":"http://dl.acm.org/citation.cfm?id=2638783","abstract":{"ja":"This paper explores if implicit gaze based annotations can support reading comprehension tasks of second language learners. We show how to use eye tracking to add implicit annotations to the text the user reads and we start by annotating physical features (reading speed, re-reading, number of fixation areas) to documents using eye tracking.\n\nWe show initial results of an ongoing experiment. So far, we recorded the eye gaze of 2 students for 2 documents. We gather initial feedback by presenting the annotated documents to two English teachers.\n\nOverall, we believe implicit annotations can be a useful feedback mechanism for second language learners."}},{"key":"Tanaka2014_UbiComp","id":"1424","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Katsuma Tanaka","en":"Katsuma Tanaka"},{"ja":"Kai Kunze","en":"Kai Kunze"},{"ja":"Motoi Iwata","en":"Motoi Iwata"},{"ja":"Koichi Kise","en":"Koichi Kise"}],"title":{"ja":"Memory Specs — An Annotation System on Google Glass using Document Image Retrieval","en":"Memory Specs — An Annotation System on Google Glass using Document Image Retrieval"},"journal":{"ja":"Proceedings of the 2014 ACM Conference on pervasive and ubiquitous computing adjunct publication (UbiComp2014)","en":"Proceedings of the 2014 ACM Conference on pervasive and ubiquitous computing adjunct publication (UbiComp2014)"},"pages":"267-270","year":2014,"month":9,"location":"Seattle, USA","review":"reviewed","presen_type":"poster","fields":["document-image-analysis"],"doi":"10.1145/2638728.2638775","abstract":{"ja":"We present a system working on a wearable computer that can annotate and retrieve annotations for signs, posters, public displays etc. The only limitation of the system: the annotated objects need to contain at least 3-5 lines of text with fixed layout. We evaluate our system in a poster session scenario with 5 use cases: retrieving annotations, adding notes, taking pictures and recording audio and attaching them to posters. As a ubiquitous note taking and annotation application, we believe the system can bridge online and offline discussions and give the user a way to reflect on seen information by providing them a web interface to all the annotations they took together with the source material (in our case poster and paper publications), helping to organize their thoughts and improving their recall.","en":"We present a system working on a wearable computer that can annotate and retrieve annotations for signs, posters, public displays etc. The only limitation of the system: the annotated objects need to contain at least 3-5 lines of text with fixed layout. We evaluate our system in a poster session scenario with 5 use cases: retrieving annotations, adding notes, taking pictures and recording audio and attaching them to posters. As a ubiquitous note taking and annotation application, we believe the system can bridge online and offline discussions and give the user a way to reflect on seen information by providing them a web interface to all the annotations they took together with the source material (in our case poster and paper publications), helping to organize their thoughts and improving their recall."}},{"key":"Utsumi2014_FFER","id":"1458","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Yuzuko Utsumi","en":""},{"ja":"Yuji Sakano","en":""},{"ja":"Keisuke Maekawa","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Scalable Face Retrieval by Simple Classifiers and Voting Scheme"},"journal":{"ja":"Proc. 1st International Workshop on Face and Facial Expression Recognition from Real World Videos (FFER 2014)"},"book_title":"Lecture Notes in Computer Science","volume":"8912","pages":"99-108","year":2014,"month":8,"publisher":"Springer","location":"Stockholm, Sweden","review":"reviewed","fields":[],"doi":"10.1007/978-3-319-13737-7_9","abstract":{"ja":"In this paper, we propose a scalable face retrieval method on large data. In order to search a particular person from videos on the Web, face recognition is one of the most effective methods. Needless to say that retrieving faces from videos are more challenging than that from a still image due to inconsistency in imaging conditions such as change of view point, lighting condition and resolution. However, dealing with them is not enough to realize the retrieval on large data. In addition, a face recognition method on the videos should be highly scalable as the number of the videos on the Web is enormous. Existing face recognition methods do not scale. In order to realize scalable face recognition, we propose a novel face recognition method. The proposed method is scalable even if the data is million-scale with high accuracy. The proposed method uses local features for face representation, and an approximate nearest neighbor (ANN) search for feature matching to reduce computational time. A voting scheme is used for recognition to compensate for low accuracy of the ANN search. We created a 5 million database and evaluated the proposed method. As results, the proposed method showed more than thousand times faster than a conventional sublinear method. Moreover, the proposed method recognized face images with a top 1000 cumulative accuracy of 100% in 139 ms recognition time (excluding preprocessing and feature extraction for the query image) per query image on the 5 million face database."}},{"key":"He2014","id":"1455","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"","en":"Jiyin He"},{"ja":"","en":"Kai Kunze"},{"ja":"","en":"Christoph Lofi"},{"ja":"","en":"Sanjay Kumar Madria"},{"ja":"","en":"Stephan Sigg"}],"title":{"en":"Towards Mobile Sensor—Aware Crowdsourcing: Architecture, Opportunitiesand Challenges"},"journal":{"en":"DASFAA Workshops 2014"},"pages":"403-412","year":2014,"month":4,"location":"Bali, Indonesia","review":"reviewed","presen_type":"oral","fields":[],"doi":"10.1007/978-3-662-43984-5_31"},{"key":"Delalandre2014","id":"1403","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"","en":"Mathieu Delalandre"},{"ja":"","en":"Motoi Iwata"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Fast and Optimal Binary Template Matching Application to Manga Copyright Protection"},"journal":{"en":"Proceedings of the 11th IAPR International Workshop on Document Analysis Systems (DAS2014)"},"pages":"298-303","year":2014,"month":4,"review":"reviewed","fields":[]},{"key":"Iwata2014_DAS","id":"1402","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"","en":"Motoi Iwata"},{"ja":"","en":"Atsushi Ito"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"A study to achieve manga character retrieval method for manga images"},"journal":{"en":"Proceedings of the 11th IAPR International Workshop on Document Analysis Systems (DAS2014)"},"pages":"309-313","year":2014,"month":4,"review":"reviewed","fields":[]},{"key":"Matsuda2014_DAS","id":"1382","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Takahiro Matsuda","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Performance Improvement in Local Feature Based Camera-Captured Character Recognition"},"journal":{"ja":"Proceedings of the 11th IAPR International Workshop on Document Analysis Systems (DAS2014)"},"pages":"196-201","year":2014,"month":4,"location":"Tours, France","review":"reviewed","presen_type":"oral","fields":["pattern-recognition"],"doi":"10.1109/DAS.2014.78"},{"key":"Cheng2014","id":"1398","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Jingyuan Cheng","en":""},{"ja":"Ayano Okoso","en":""},{"ja":"Kai Kunze","en":""},{"ja":"Niels Henze","en":""},{"ja":"Albrecht Schmidt","en":""},{"ja":"Paul Lukowicz","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"On the Tip of my Tongue ? A Non-Invasive Pressure-BasedTongue Interface"},"journal":{"ja":"Proc. 5th Augmented Human (AH2014)"},"year":2014,"month":3,"review":"reviewed","fields":[]},{"key":"Iwamura2014_AH","id":"1389","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Kai Kunze","en":""},{"ja":"Yuya Kato","en":""},{"ja":"Yuzuko Utsumi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Haven't we met before? — A Realistic Memory Assistance System to Remind You of The Person in Front of You"},"journal":{"ja":"Proc. 5th Augmented Human (AH2014)"},"num_pages":"4","year":2014,"month":3,"location":"Kobe, Japan","review":"reviewed","fields":[],"doi":"10.1145/2582051.2582083","abstract":{"ja":"This paper presents a perceived real-time system for memory augmentation.\nWe propose a realistic approach to realize a memory assistance system,\nfocusing on retrieving the person in front of you.\nThe proposed system is capable of fully automatic indexing and\nis scalable in the database size. \nWe utilize face recognition to show the user previous encounters with\nthe person they are currently looking at.\nThe system works fast (under 200 ms, perceived real time) with a\ndecent database size (45 videos of 24 people).\nWe also provide evidence in terms of an online questionnaire that\nour proposed memory augmentation system is useful and would be worn \nby most of the participants if it can be implemented in an unobtrusive way."}},{"key":"Ishimaru2014","id":"1399","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Shoya Ishimaru","en":""},{"ja":"Jens Weppner","en":""},{"ja":"Kai Kunze","en":""},{"ja":"Andreas Bulling","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"Paul Lukowicz","en":""}],"title":{"ja":"In the Blink of an Eye — Combining Head Motion and Eye Blink Frequency for Activity Recognition with Google Glass."},"journal":{"ja":"Proceedings of the 5th Augmented Human International Conference"},"pages":"150-153","year":2014,"month":3,"review":"reviewed","fields":[],"abstract":{"ja":"We demonstrate how information about eye blink frequency and head motion patterns derived from Google Glass sensors can be used to distinguish different types of high level activities. While it is well known that eye blink frequency is correlated with user activity, our aim is to show that (1) eye blink frequency data from an unobtrusive, commercial platform which is not a dedicated eye tracker is good enough to be useful and (2) that adding head motion patterns information significantly improves the recognition rates. The method is evaluated on a data set from an experiment containing five activity classes (reading, talking, watching TV, mathematical problem solving, and sawing) of eight participants showing 67% recognition accuracy for eye blinking only and 82% when extended with head motion patterns."}},{"key":"Toyama2014","id":"1379","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Takumi Toyama","en":""},{"ja":"Daniel Sonntag","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"Takahiro Matsuda","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"A Mixed Reality Head-Mounted Text Translation System Using Eye Gaze Input"},"journal":{"ja":"Proc. 2014 International Conference on Intelligent User Interfaces (IUI'14)"},"pages":"329-334","year":2014,"month":2,"location":"Haifa, Israel","review":"reviewed","fields":[],"doi":"10.1145/2557500.2557528","abstract":{"ja":"Efficient text recognition has recently been a challenge for augmented reality systems. In this paper, we propose a system with the ability to provide translations to the user in realtime. We use eye gaze for more intuitive and efficient input for ubiquitous text reading and translation in head mounted displays (HMDs). The eyes can be used to indicate regions of interest in text documents and activate optical-characterrecognition (OCR) and translation functions. Visual feedback and navigation help in the interaction process, and text snippets with translations from Japanese to English text snippets, are presented in a see-through HMD. We focus on travelers who go to Japan and need to read signs and propose two different gaze gestures for activating the OCR text reading and translation function. We evaluate which type of gesture suits our OCR scenario best. We also show that our gaze-based OCR method on the extracted gaze regions provide faster access times to information than traditional OCR approaches. Other benefits include that visual feedback of the extracted text region can be given in real-time, the Japanese to English translation can be presented in real-time, and the augmentation of the synchronized and calibrated HMD in this mixed reality application are presented at exact locations in the augmented user view to allow for dynamic text translation management in head-up display systems."}},{"key":"Utsumi2014","id":"1380","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Yuzuko Utsumi","en":""},{"ja":"Yuki Shiga","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Kai Kunze","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Document Type Classification Toward Understanding Reading Habits"},"journal":{"ja":"Proceedings of 20th Korea?Japan Joint Workshop on Frontiers of Computer Vision"},"presen_id":"3","pages":"11-17","year":2014,"month":2,"location":"Okinawa, Japan","review":"reviewed","fields":["document-image-analysis"]},{"key":"Matsuda2014_PRMU","id":"1431","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"松田 崇宏","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"データクラスタリングによる局所特徴ベース情景内文字認識手法の改善"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"114","number":"356","presen_id":"PRMU2014-75","pages":"57-62","year":2014,"month":12,"location":"福岡県福岡市","review":"not_reviewed","presen_type":"oral","fields":["pattern-recognition"]},{"key":"Kazuyo2014","id":"1428","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"吉村 和代","en":""},{"ja":"Kai Kunze","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"fNIRSで取得した脳活性に基づく文書理解度推定法"},"book_title":"電子情報通信学会技術研究報告","presen_id":"MBE2014-65","pages":"35-40","year":2014,"month":11,"location":"宮城県仙台市","review":"not_reviewed","presen_type":"oral","fields":[]},{"key":"Ishimaru2014_PRMU","id":"1425","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"石丸 翔也","en":"Shoya Ishimaru"},{"ja":"クンツェ カイ","en":"Kai Kunze"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"眼電位を用いたみんなのための眼球運動認識","en":"Toward Eye Movement Recognition in the Large using Electrooculography"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"114","number":"230","presen_id":"PRMU2014-62","pages":"45-48","year":2014,"month":10,"location":"千葉県千葉市","review":"not_reviewed","presen_type":"oral","fields":["pattern-recognition"],"abstract":{"ja":"人の認知的な行動を認識するアプローチのひとつとして，眼球運動の記録がある．記録には，アイトラッカと呼ばれる専用の装置が必要とするため，利用環境が限られるという課題がある．この問題を解決するため，本研究では眼電位センサを搭載した安価なスマートアイウエア (J!NS MEME) によって，誰もが手軽に眼球運動を認識できるシステムの実現を目指す．\n本稿では実現のための第一歩として，眼電位センサを観察して瞬きや視線方向の認識に有用な特徴を探す．次に，眼電位と加速度データを用いて認知的な行動を含む様々な行動(読書, タイピング, 食事, 会話)を識別する．被験者2人4つの行動の識別に対して，6秒間のフレーム毎の識別で70%の識別率を得た．1分間に含まれるフレームの投票処理では100％の識別率を得ることができた．","en":"In this report, we discuss how to detect eye movement and blinks robustly using consumer smart glasses called JINS MEME for the use in cognitive activity recognition. We are able to distinguish of 4 activities (reading, typing, eating and talking) for 2 users using the sensor data (EOG and acceleration) from the glasses with an accuracy of 70% for 6 seconds windows and up to 100% for a 1 minute majority decision."}},{"key":"Utsumi2014_CVIM","id":"1419","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"内海ゆづ子","en":"Yuzuko Utsumi"},{"ja":"坂野悠司","en":"Yuji Sakano"},{"ja":"前川敬介","en":"Keisuke Maekawa"},{"ja":"岩村雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬浩一","en":"Koichi Kise"}],"title":{"ja":"大規模データベースに対する高速な顔認識"},"journal":{"ja":"情報処理学会研究報告コンピュータビジョンとイメージメディア（CVIM）"},"volume":"2014-CVIM-193","number":"12","pages":"1","year":2014,"month":9,"review":"not_reviewed","fields":["pattern-recognition"],"note":{"ja":"ここに掲載した著作物の利用に関する注意 本著作物の著作権は情報処理学会に帰属します。本著作物は著作権者である情報処理学会の許可のもとに掲載するものです。ご利用に当たっては「著作権法」ならびに「情報処理学会倫理綱領」に従うことをお願いいたします。","en":"Notice for the use of this material The copyright of this material is retained by the Information Processing Society of Japan (IPSJ). This material is published on this web site with the agreement of the author (s) and the IPSJ. Please be complied with Copyright Law of Japan and the Code of Ethics of the IPSJ if any users wish to reproduce, make derivative work, distribute or make available to the public any part or whole thereof.\nAll Rights Reserved, Copyright (C) Information Processing Society of Japan.\nComments are welcome. Mail to address editj＠ipsj.or.jp, please."}},{"key":"Miyazawa2014_MIRU","id":"1414","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"宮沢 嘉希","en":""},{"ja":"岩田 基","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"早期棄却と早期認識を用いた文書画像検索の提案"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2014)"},"presen_id":"SS2-24","year":2014,"month":7,"location":"岡山県岡山市","review":"not_reviewed","presen_type":"poster","fields":["document-image-analysis","document-image-retrieval"]},{"key":"Tanaka2014_MIRU","id":"1412","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"田中 克磨","en":""},{"ja":"カイ クンツェ","en":""},{"ja":"岩田 基","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"Google Glass て?動作するホ?スターアノテーションシステム"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2014)"},"presen_id":"DS2-9","year":2014,"month":7,"location":"岡山県岡山市","review":"not_reviewed","presen_type":"poster","fields":["document-image-analysis","document-image-retrieval"]},{"key":"Ishimaru2014_DSa","id":"1416","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"石丸 翔也","en":"Shoya Ishimaru"},{"ja":"Jens Weppner","en":"Jens Weppner"},{"ja":"Kai Kunze","en":"Kai Kunze"},{"ja":"Paul Lukowicz","en":"Paul Lukowicz"},{"ja":"Andreas Dengel","en":"Andreas Dengel"},{"ja":"Andreas Bulling","en":"Andreas Bulling"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"Google Glassを用いた瞬き検出と認知的行動の認識"},"journal":{"ja":"第17回画像の認識・理解シンポジウム予稿集"},"presen_id":"DS2-10","year":2014,"month":7,"location":"岡山県岡山市","review":"not_reviewed","presen_type":"poster","fields":[]},{"key":"Ishimaru2014_DSb","id":"1417","type":"domestic_conf","language":"EN","show":"show","authors":[{"ja":"Shoya Ishimaru","en":""},{"ja":"Kai Kunze","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Yuji Uema","en":""},{"ja":"Masahiko Inami","en":""}],"title":{"ja":"J!NS MEME — Activity Recognition using Smart Eyewear"},"journal":{"ja":"第17回画像の認識・理解シンポジウム予稿集"},"presen_id":"DS2-8","year":2014,"month":7,"location":"岡山県岡山市","review":"not_reviewed","presen_type":"poster","fields":[]},{"key":"Utsumi2014_SS","id":"1415","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"内海ゆづ子","en":"Yuzuko Utsumi"},{"ja":"森俊介","en":"Shunsuke Mori"},{"ja":"岩村雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬浩一","en":"Koichi Kise"}],"title":{"ja":"顔の個人性を考慮した特徴点検出"},"journal":{"ja":"第17回画像の認識・理解シンポジウム予稿集"},"presen_id":"SS3-2","year":2014,"month":7,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Iwata2014","id":"1457","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"岩田基","en":"Motoi Iwata"}],"title":{"ja":"電子透かし技術を利用した印刷物やサイネージ映像からの付加情報の抽出","en":"Extracting Additional Information from Printed Materials and Videos on Signage by Utilizing Digital Watermarking"},"journal":{"ja":"Imaging Conference Japan 2014 論文集"},"pages":"103-106","year":2014,"month":6,"publisher":"日本画像学会","review":"not_reviewed","fields":[]},{"key":"Mori2014_CVIM","id":"1406","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"森 俊介","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"顔の個人性を反映した複数の顔モデルによる特徴点検出精度の向上"},"journal":{"ja":"情報処理学会研究報告コンピュータビジョンとイメージメディア(CVIM)"},"book_title":"情報処理学会研究報告","volume":"55","number":"10","presen_id":"CVIM2014-11","year":2014,"month":5,"review":"not_reviewed","presen_type":"oral_and_poster","fields":[],"note":{"ja":"ここに掲載した著作物の利用に関する注意 本著作物の著作権は情報処理学会に帰属します。本著作物は著作権者である情報処理学会の許可のもとに掲載するものです。ご利用に当たっては「著作権法」ならびに「情報処理学会倫理綱領」に従うことをお願いいたします。","en":"Notice for the use of this material The copyright of this material is retained by the Information Processing Society of Japan (IPSJ). This material is published on this web site with the agreement of the author (s) and the IPSJ. Please be complied with Copyright Law of Japan and the Code of Ethics of the IPSJ if any users wish to reproduce, make derivative work, distribute or make available to the public any part or whole thereof.\nAll Rights Reserved, Copyright (C) Information Processing Society of Japan.\nComments are welcome. Mail to address editj＠ipsj.or.jp, please."},"abstract":{"ja":"顔の特徴点を検出する方法として，顔モデルを用いる方法がある．これは与えられた顔画像に対して，顔モデルをフィッティングさせて顔の形状を推定することで，顔の特徴点を検出する手法である．これまで提案されてきた顔モデルベースの特徴点検出手法では，複数の人物の顔画像を用いて1 つの顔モデルを構築していた．しかし，人によって顔の特徴は異なるため，1 つのモデルでは顔の個人性に対応することが難しく，検出した特徴点の誤差が大きくなることがある．そこで本研究では，顔の特徴点検出精度の向上を目的と，複数の顔モデルを構築して特徴点を検出する手法を提案する．提案手法では複数の顔モデルがあり，それぞれの顔モデルで用いられる学習画像は異なる人物の顔画像である．そのため各顔モデルは顔の個人性を反映しており，複数の顔モデルをフィッティングに用いることで顔の個人性に対応可能となる．また最も適合した顔モデルから，与えられた顔画像がどのような顔と類似するか大分類することが可能となる．実験の結果，1 人につき1 つの顔モデルを構築することで，特徴点検出精度は1.3[pixel]向上し，分類の精度は96:0% となった．"}},{"key":"Shiga2014_CVIM","id":"1409","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"志賀 優毅","en":"Yuki Shiga"},{"ja":"内海 ゆづ子","en":"Yuzuko Utsumi"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"カイ クンツェ","en":"Kai Kunze"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"一人称視点画像を用いた文書画像の分類"},"journal":{"ja":"情報処理学会研究報告コンピュータビジョンとイメージメディア(CVIM)"},"volume":"15","pages":"1-7","year":2014,"month":5,"publisher":"一般社団法人情報処理学会","location":"大阪府東大阪市","review":"not_reviewed","presen_type":"oral_and_poster","fields":["document-image-analysis","pattern-recognition"],"note":{"ja":"ここに掲載した著作物の利用に関する注意 本著作物の著作権は情報処理学会に帰属します。本著作物は著作権者である情報処理学会の許可のもとに掲載するものです。ご利用に当たっては「著作権法」ならびに「情報処理学会倫理綱領」に従うことをお願いいたします。","en":"Notice for the use of this material The copyright of this material is retained by the Information Processing Society of Japan (IPSJ). This material is published on this web site with the agreement of the author (s) and the IPSJ. Please be complied with Copyright Law of Japan and the Code of Ethics of the IPSJ if any users wish to reproduce, make derivative work, distribute or make available to the public any part or whole thereof.\nAll Rights Reserved, Copyright (C) Information Processing Society of Japan.\nComments are welcome. Mail to address editj＠ipsj.or.jp, please."},"abstract":{"ja":"読書の時間，頻度，あるいはどのような種類の本を読んでいるかなどの情報を把握することは，自己管理や学習到達度の目安として役立つため大変重要であると考えられている．しかし，それらの情報の手動記録には多くの時間を消費してしまう．読書習慣の情報を自動的に記録，分析するために，我々はウェアラブルカメラで撮影された一人称視点の文書画像を分類し，ユーザが読んでいる文書を推定する問題を提案する．文書画像の分類は一種の一般物体認識といえるため，一般物体認識で広く用いられている特徴量を使用し，データベースに登録されていない文書画像を各クラスに分類する．Bag-of-Featuresを用いた手法で実験した結果，81\\%の正解率を得ることができた．"}},{"key":"Miyazawa2014_TR","id":"1408","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"宮沢 嘉希","en":""},{"ja":"岩田 基","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"早期棄却と早期認識による文書画像検索の高速化"},"journal":{"ja":"電子情報通信学会技術研究報告（原稿の締切に間に合わなかったため，掲載されていません．）"},"year":2014,"month":5,"location":"愛知県名古屋市","review":"not_reviewed","presen_type":"oral","fields":["document-image-analysis","information-retrieval","pattern-recognition","object-recognition"],"abstract":{"ja":"高速な文書画像検索手法の一つに，LLAH がある．LLAH は高速・高精度な文書画像検索を実現できる一\n方，図表やグラフを多く含む文書を対象としたとき，精度が著しく低下するという欠点がある．一方，幅広い対象に\n適用可能な画像検索手法として，SIFT などの局所特徴量を用いた手法があるが，低速であるという欠点がある．本\n稿では，局所特徴量を用いた画像検索手法に，早期認識と早期棄却の二つの概念を導入することで，幅広い対象に適\n用可能で，なおかつ高速・高精度な文書画像検索手法を提案する．早期認識は，計算コストの少ない処理を繰り返し，\n検索に十分な証拠が得られた時点で処理を終了することで達成する．早期棄却は，少ない計算コストの処理から，検\n索に有効ではない特徴抽出の範囲を決定し，それを破棄することで実現する．実験の結果，1 万ページのデータベー\nスに対して，精度94%，処理時間45.6[ms] で検索可能であることが分かった．"}},{"key":"Suzuki2014_PRMU","id":"1391","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"鈴木若菜","en":""},{"ja":"外山託海","en":""},{"ja":"Kai Kunze","en":""},{"ja":"岩村雅一","en":""},{"ja":"黄瀬浩一","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"Andreas Bulling","en":""}],"title":{"ja":"読書アノテーション支援システムに向けたアイジェスチャの提案"},"journal":{"ja":"電子情報通信学会技術報告"},"volume":"113","number":"493","presen_id":"PRMU2013-182","pages":"85-90","year":2014,"month":3,"review":"not_reviewed","fields":[]},{"key":"Eguchi2014","id":"1396","type":"domestic_conf","language":"JA","show":"show_except_abstract","authors":[{"ja":"江口 健太","en":"Kenta EGUCHI"},{"ja":"岩田 基","en":"Motoi IWATA"},{"ja":"黄瀬 浩一","en":"Koichi KISE"}],"title":{"ja":"均等色空間の性質を利用したJPEG画像を対象とする電子透かし法","en":"Digital watermarking for JPEG images using uniform color space"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"113","number":"480","presen_id":"EMM2013-112","pages":"75-80","year":2014,"month":3,"review":"not_reviewed","fields":[]},{"key":"Kato2014_PRMU","id":"1388","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"加藤 祐也","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"Kai Kunze","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"顔画像検索を用いたリアルタイム人物記憶補助システム"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"113","number":"493","presen_id":"PRMU2013-203","pages":"209-214","year":2014,"month":3,"review":"not_reviewed","fields":[]},{"key":"Kawashi2014_PRMU","id":"1385","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"川市 仁史","en":""},{"ja":"Kai Kunze","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"Reading-Life Logの実現に向けた万語計の提案"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"113","number":"493","presen_id":"PRMU2013-181","pages":"79-84","year":2014,"month":3,"review":"not_reviewed","fields":["document-image-analysis"],"abstract":{"ja":"Reading-Life Log とは，文書に対する「読む」という行動の記録である．これを解析することでユーザの趣向や考えを把握し，適した情報を提供することが可能になると考えられる．最もシンプルなReading-Life Logの一つとして，ユーザの読んだ単語数を記録することが挙げられる．我々は，ユーザの読んだ単語数を記録するものを万語計と呼ぶ．万語計では，ユーザの眼の動きや読んでいる文書の情報を取得することにより，読んだ単語数を推定する．本稿では，4つの万語計について比較検討する．具体的には，(1) 単純に読書時間から推定する手法，(2) 読んだ文書の一行あたりの平均単語数と読んだ行数から推定する手法，(3) (2)の手法に視点情報を加味した手法，(4) 視点情報のみを用いた手法\nである．実験の結果，最も精度が高い手法は(3)であり，その誤差が11.3%であることがわかった．"}},{"key":"Matozaki2014_PRMU","id":"1383","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"的崎 伸彰","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"近似逆最近傍探索によるBM25を用いた映像検索の高速化"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"113","number":"493","presen_id":"PRMU2013-183","pages":"91-96","year":2014,"month":3,"fields":["pattern-recognition","object-recognition","nearest-neighbor-search"]},{"key":"Sato2014_PRMU","id":"1386","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"佐藤 瞳","en":""},{"ja":"松田 崇宏","en":""},{"ja":"池田 佑輝","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"大規模日本語情景文字データセットOPU-JST-1"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"113","number":"493","presen_id":"PRMU2013-176","pages":"49-54","year":2014,"month":3,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"スマートフォンやタブレットなどの小型端末の普及が進んでいることから，それらを用いた様々な情景中\nの文字を認識するサービスやアプリケーションの需要が高まっている．情景中の文字は多種多様な形状を持ち，撮影\n時に射影歪みや照明環境の影響を受ける．解像度が低下する場合もある．そのため，これらの文字の検出・認識は，ス\nキャナで撮像した場合など，撮像環境をコントロールした場合よりも困難である．この問題に対処する有効な方法と考\nえられているのが，種々の劣化を受けた文字データを大量に集めて，識別器の学習に使用することである．本研究で\nは全方位カメラを用いて商店街などを撮影し，得られた画像からなる情景中の日本語文字データセット，OPU-JST-1\nを構築したので，それを報告する．このデータセットは，時系列情報を持つ情景画像31,410 枚と，790,257 個の文字\n領域のラベル，2,764,230 文字を収録している．"}},{"key":"Tanaka2014_PRMU","id":"1400","type":"domestic_conf","language":"JA","show":"show_except_abstract","authors":[{"ja":"田中 克磨","en":"Katsuma Tanaka"},{"ja":"カイ クンツェ","en":"Kai Kunze"},{"ja":"岩田 基","en":"Motoi Iwata"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"ポスター発表のためのアノテーション共有システム","en":"A Digital Annotation Sharing System for Poster Sessions"},"book_title":"電子情報通信学会技術研究報告","volume":"113","number":"431","presen_id":"PRMU2013-136","pages":"83-88","year":2014,"month":2,"review":"not_reviewed","fields":["document-image-retrieval"]},{"key":"Bouclans2014","id":"1394","type":"domestic_conf","language":"EN","show":"show_except_abstract","authors":[{"ja":"Olivier Le Bas de Bouclans","en":""},{"ja":"Kai Kunze","en":""},{"ja":"Hiroki Fujiyoshi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Mental Task Classification Using Electroencephalogram"},"journal":{"ja":"IEICE Technical Report"},"volume":"113","number":"431","presen_id":"PRMU2013-155","pages":"129-134","year":2014,"month":2,"review":"not_reviewed","fields":[]},{"key":"Akinishi2014_Symp","id":"1381","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"穐西 俊弥","en":"Toshiya Akinishi"},{"ja":"岩田 基","en":"Motoi Iwata"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"再撮影動画から透かし入り動画の領域を推定する電子透かし法","en":"Digital Watermarking Method Estimating Watermarked Video Region from Recaptured Video"},"journal":{"ja":"2014年暗号と情報セキュリティシンポジウム概要集"},"year":2014,"month":1,"review":"not_reviewed","fields":[],"abstract":{"ja":"動画に対して電子透かしを用い，人に知覚されないように付加情報を埋め込み，それをカメラで再撮影することで情報を取り出す手法について検討する．透かし入り動画にスマートフォンをかざすことで関連情報にアクセスできるようにすることが目的である．透かし入り動画を再撮影した動画には，透かし抽出の対象である透かし入り動画の領域以外に背景領域も含まれる．そこで，電子透かし法を，再撮影動画内の透かし入り動画の領域を推定する処理と透かしの埋め込み・抽出を行う処理に分けて考える．まず，透かしの埋め込み・抽出にはフレーム間の差分を利用する手法を用いる．提案手法は携帯端末上でのリアルタイム処理を目指しているため，小さい計算コストが望まれる．そこで，周波数変換などと比べ計算コストが小さい単純な和算，減算のみを用いて計算できるフレーム間差分を利用する方法を用いた．次に，透かし入り動画の領域推定には，透かしとは別に位置ずれ同期用の情報を埋め込むことも考えられるが，画質劣化の原因になるため好ましくない．そこで本稿では，透かし自体の痕跡を利用して透かしを抽出する方法を新たに提案する．実験では，スマートフォンにより撮影した，手ぶれを含む再撮影動画から透かしを抽出し，提案手法の性能を評価する．"}},{"key":"Kise2014","id":"1456","type":"book","language":"EN","show":"show_except_datafile","authors":[{"ja":"Koichi Kise","en":""}],"title":{"ja":"Page Segmentation Techniques in Document Analysis"},"book_title":"Handbook of Document Image Processing and Recognition","year":2014,"month":10,"publisher":"Springer","review":"not_reviewed","fields":["document-image-analysis"],"doi":"10.1007/978-0-85729-859-1_5"},{"key":"Iwamura2014_LNCS","id":"1405","type":"book","language":"EN","show":"show_except_datafile","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Faisal Shafait (Eds.)","en":""}],"title":{},"book_title":"Camera-Based Document Analysis and Recognition - 5th International Workshop, CBDAR 2013, Revised Selected Papers","volume":"8357","presen_id":"LNCS","year":2014,"month":4,"publisher":"Springer","review":"not_reviewed","fields":[],"doi":"10.1007/978-3-319-05167-3","url":"http://www.springer.com/computer/image+processing/book/978-3-319-05166-6","abstract":{"ja":"This book constitutes the thoroughly refereed post-workshop proceedings of the 5th International Workshop on Camera-Based Document Analysis and Recognition, CBDAR 2013, held in Washington, DC, USA, in August 2013. The 14 revised full papers presented were carefully selected during two rounds of reviewing and improvement from numerous original submissions. Intended to give a snapshot of the state-of-the-art research in the field of camera based document analysis and recognition, the papers are organized in topical sections on text detection and recognition in scene images, and camera-based systems."}},{"key":"Kise2014_HDIP","id":"1454","type":"book","language":"JA","show":"show_except_datafile","authors":[{"ja":"","en":"Koichi Kise"}],"title":{"en":"Page Segmentation Techniques in Document Analysis"},"journal":{"en":"Handbook of Document Image Processing and Recognition(Eds. by D.Doermann and K.Tombre)"},"pages":"135-175","year":2014,"publisher":"Springer―Verlag","address":"London","review":"not_reviewed","fields":[]},{"key":"Imura2014","id":"1393","type":"master_thesis","language":"JA","show":"show_except_abstract","authors":[{"ja":"伊村 政志","en":""}],"title":{"ja":"ピントのずれに頑健なパターン認識"},"year":2014,"month":3,"fields":["pattern-recognition","computational-photography"]},{"key":"Kato2014_MThesis","id":"1395","type":"master_thesis","language":"JA","show":"show_except_abstract","authors":[{"ja":"加藤 祐也","en":""}],"title":{"ja":"顔画像検索を用いたリアルタイム人物記憶補助システム"},"year":2014,"month":3,"review":"not_reviewed","fields":[]},{"key":"Kawashi2014_MThesis","id":"1770","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"川市 仁史","en":""}],"title":{"ja":"Wordometer によるReading-Life Log の実現","en":"Implementation of Reading-Life Log by Wordometer"},"year":2014,"month":3,"fields":[]},{"key":"Matozaki2014_MThesis","id":"1384","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"的崎 伸彰","en":"Nobuaki Matozaki"}],"title":{"ja":"BM25を用いた映像認識の近似逆最近傍探索による高速化","en":"Speed-up of video recognition with BM25 by approximate reverse nearest neighbor search"},"year":2014,"month":3,"review":"not_reviewed","fields":["pattern-recognition","object-recognition"]},{"key":"Olivier2014","id":"1772","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"LE BAS DE BOUCLANS Olivier","en":"LE BAS DE BOUCLANS Olivier"}],"title":{"ja":"脳波を用いたメンタルタスクの分類","en":"Mental Task Classi\fcation UsingElectroencephalogram"},"year":2014,"month":3,"fields":[]},{"key":"Suzuki2014_MThesis","id":"1771","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"鈴木 若菜","en":""}],"title":{"ja":"読書アノテーションのためのアイジェスチャの認識","en":"Recognition of Eye Gestures for Annotation of Reading Activities"},"year":2014,"month":3,"fields":[]},{"key":"Ishimaru2014_BThesis","id":"1401","type":"bachelors_thesis","language":"EN","show":"show_except_datafile","authors":[{"ja":"Shoya Ishimaru","en":""}],"title":{"ja":"Activity Recognition with Google Glass: Combining Head Motion and Eye Blink Frequency"},"year":2014,"month":3,"review":"not_reviewed","fields":[],"abstract":{"ja":"This thesis demonstrates how information about eye blink frequency and head motion patterns derived from Google Glass sensors can be used to distinguish different types of high level activities. While it is well known that eye movement is correlated with user activities, the aim of this research is to show that (1) eye blink frequency data from an unobtrusive, commercial platform which is not a dedicated eye tracker is good enough to be useful and (2) that adding head motion patterns information significantly improves the recognition rates. The method is evaluated on a data set containing five activity classes (reading a book, watching a video, solving mathematical tasks, sawing a cardboard and talking) of eight participants showing 67% recognition accuracy for eye blink only and 82% when extended with head motion patterns."}},{"key":"Murata2013","id":"1468","type":"journal_paper","language":"JA","show":"show_except_datafile","authors":[{"ja":"村田 晴美","en":"Harumi MURATA"},{"ja":"荻原 昭夫","en":"Akio OGIHARA"},{"ja":"岩田 基","en":"Motoi IWATA"},{"ja":"汐崎 陽","en":"Akira SHIOZAKI"}],"title":{"ja":"単一楽器演奏曲を対象とするサンプリング音を用いた音楽電子透かし法","en":"An Audio Watermarking Method Using Sampled Sound for Single Instrumental Music"},"journal":{"ja":"電子情報通信学会論文誌","en":"IEICE Trans."},"volume":"J96-D","number":"4","pages":"941-951","year":2013,"month":4,"review":"reviewed","fields":[]},{"key":"Sun2013_IJDAR","id":"1310","type":"journal_paper","language":"EN","show":"show_except_datafile","authors":[{"ja":"Weihan Sun","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Detection of exact and similar partial copies for copyright protection of manga"},"journal":{"ja":"International Journal on Document Analysis and Recognition (IJDAR)"},"volume":"16","number":"1","pages":"1-19","year":2013,"month":3,"review":"reviewed","fields":["document-image-analysis","information-retrieval","pattern-recognition"],"doi":"10.1007/s10032-013-0199-y","abstract":{"ja":"Manga, a kind of Japanese comic book, is an important genre in the realm of image publications requiring copyright protection. To copy manga, illegal users generally focus on certain interesting parts from which to make partial copies to apply in their own drawings. With respect to their sources, copying of manga can be divided into two types: (1) exact copies, which duplicate specific contents of manga, such as scanned manga publications (printed copies) and traced outlines of manga (hand-drawn copies), and (2) similar partial copies, which infringe the copyright of manga characters based on their features. In this paper, we propose applying content-based image retrieval methods to detect both exact and similar copies based on two kinds of regions of interest (ROIs): generic ROIs and face ROIs. The method is able not only to locate the partial copies from images with complex backgrounds, but also to report the corresponding copied parts of copyrighted manga pages for exact copy detection and copied manga characters for similar copy detection. The experimental results prove high performance of the proposed method for detecting printed partial copies. In addition, 85 % of hand-drawn and 77 % of similar partial copies were detected with relatively high precision using a database containing more than 10,000 manga pages."}},{"key":"Iwamura2013_ICCV","id":"1357","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Tomokazu Sato","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"What Is the Most Efficient Way to Select Nearest Neighbor Candidates for Fast Approximate Nearest Neighbor Search?"},"journal":{"ja":"Proc. 14th International Conference on Computer Vision (ICCV 2013)"},"pages":"3535-3542","year":2013,"month":12,"location":"Sydney, Australia","review":"reviewed","presen_type":"poster","fields":["nearest-neighbor-search"],"doi":"10.1109/ICCV.2013.439","note":{"ja":"<ol>\n<li>C and P in Table 1 are typos for c and p, respectively.</li>\n<li>Experimental result of the following paper [a] that shows BDH was worse than IMI is theoretically wrong because BDH is a generalization of IMI and never gets worse than IMI.<br />\n[a] Heo, J.-P.; Lin, Z.; Shen, X.; Brandt, J. & Yoon, S.-E.<br />\nShortlist Selection with Residual-Aware Distance Estimator for K-Nearest Neighbor Search<br />\nProc. CVPR, IEEE, 2016, pp.2009-2017</li>\n</ol>"}},{"key":"Ishimaru2013","id":"1686","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Shoya Ishimaru","en":""},{"ja":"Kai Kunze","en":""},{"ja":"Yuzuko Utsumi","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Where Are You Looking At? - Feature-Based Eye Tracking on Unmodified Tablets"},"journal":{"ja":"Proc. 2nd Asian Conference on Pattern Recognition (ACPR2013)"},"year":2013,"month":11,"location":"Okinawa, Japan","review":"not_reviewed","fields":[],"doi":"10.1109/ACPR.2013.190"},{"key":"Iwamura2013_ACPR","id":"1358","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Takuya Kobayashi","en":""},{"ja":"Takahiro Matsuda","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Recognition of Layout-Free Characters on Complex Background"},"journal":{"ja":"Proc. 2nd Asian Conference on Pattern Recognition (ACPR2013)"},"num_pages":"2","year":2013,"month":11,"location":"Okinawa, Japan","review":"reviewed","fields":["document-image-analysis","pattern-recognition"],"doi":"10.1109/ACPR.2013.191"},{"key":"Iwata2013","id":"1377","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"","en":"Motoi Iwata"},{"ja":"","en":"Kenji Yamanaka"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Digital Watermarking Method to Extract Watermarks from Printed Matters with Cell Phone by Using Finder Patterns and Alignment Pattern of QR Code"},"journal":{"en":"Proc. of the 2nd Asian Conference on Pattern Recognition (ACPR2013)"},"pages":"391-395","year":2013,"month":11,"review":"reviewed","fields":[]},{"key":"Tanaka2013","id":"1378","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"","en":"Kazuma Tanaka"},{"ja":"","en":"Motoi Iwata"},{"ja":"","en":"Kai Kunze"},{"ja":"","en":"Masakazu Iwamura"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Share Me ? A Digital Annotation Sharing Service for Paper Documents with Multiple Clients Support"},"journal":{"en":"Proc. of the 2nd Asian Conference on Pattern Recognition (ACPR2013)"},"pages":"779-782","year":2013,"month":11,"location":"Okinawa, Japan","review":"reviewed","fields":["document-image-retrieval"],"doi":"10.1109/ACPR.2013.182"},{"key":"Utsumi2013_IAPR","id":"1376","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"","en":"Yuzuko Utsumi"},{"ja":"","en":"Misa Katte"},{"ja":"","en":"Masakazu Iwamura"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Event Detection Based on Noisy Object Information"},"journal":{"en":"Proceedings of 2013 Second IAPR Asian Conference on Pattern Recognition"},"pages":"572-575","year":2013,"month":11,"location":"Okinawa, Japan","review":"reviewed","fields":["pattern-recognition"],"doi":"10.1109/ACPR.2013.85"},{"key":"Toyama2013b","id":"1360","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Takumi Toyama","en":""},{"ja":"Wakana Suzuki","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"User Attention Oriented Augmented Reality on Documents with DocumentDependent Dynamic Overlay"},"journal":{"ja":"Proc. ISMAR2013"},"year":2013,"month":10,"location":"Adelaide, Australia","review":"reviewed","presen_type":"poster","fields":["document-image-analysis"],"doi":"10.1109/ISMAR.2013.6671814"},{"key":"Kunze2013","id":"1368","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Kai Kunze","en":""},{"ja":"Yuzuko Utsumi","en":""},{"ja":"Yuki Shiga","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Andreas Bulling","en":""}],"title":{"ja":"I know what you are reading: recognition of document types using mobile eye tracking"},"journal":{"ja":"Proceedings of the 17th annual international symposium on International symposium on wearable computers"},"pages":"113-116","year":2013,"month":9,"location":"Zurich, Switzerland","review":"reviewed","presen_type":"oral","fields":[],"doi":"10.1145/2493988.2494354"},{"key":"Cheng2013","id":"1365","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Jingyuan Cheng","en":""},{"ja":"Bo Zhou","en":""},{"ja":"Kai Kunze","en":""},{"ja":"Carl Christian Rheinländer","en":""},{"ja":"Sebastian Wille","en":""},{"ja":"Norbert Wehn","en":""},{"ja":"Jens Weppner","en":""},{"ja":"Paul Lukowicz","en":""}],"title":{"ja":"Activity recognition and nutrition monitoring in every day situations with a textile capacitive neckband"},"journal":{"ja":"Proceedings of the 2013 ACM conference on Pervasive and ubiquitous computing adjunct publication"},"pages":"155-158","year":2013,"month":9,"review":"reviewed","fields":[]},{"key":"Kunze2013_ACMa","id":"1366","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Kai Kunze","en":""},{"ja":"Katsuma Tanaka","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Annotate me: supporting active reading using real-time document image retrieval on mobile devices"},"journal":{"ja":"Proceedings of the 2013 ACM conference on Pervasive and ubiquitous computing adjunct publication"},"pages":"231-234","year":2013,"month":9,"location":"Zurich, Switzerland","review":"reviewed","presen_type":"poster","fields":["document-image-retrieval"],"doi":"10.1145/2494091.2494165"},{"key":"Kunze2013_ACMb","id":"1367","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Kai Kunze","en":""},{"ja":"Shoya Ishimaru","en":""},{"ja":"Yuzuko Utsumi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"My reading life: towards utilizing eyetracking on unmodified tablets and phones"},"journal":{"ja":"Proceedings of the 2013 ACM conference on Pervasive and ubiquitous computing adjunct publication"},"pages":"283-286","year":2013,"month":9,"location":"Zurich, Switzerland","review":"reviewed","presen_type":"poster","fields":[],"doi":"10.1145/2494091.2494179"},{"key":"Iwamura2013_CBDAR","id":"1359","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Takuya Kobayashi","en":""},{"ja":"Takahiro Matsuda","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Fast and Layout-Free Camera-Based Character Recognition on Complex Backgrounds"},"journal":{"ja":"Handout of 5th International Workshop on Camera-Based Document Analysis and Recognition (CBDAR2013)"},"num_pages":"1","year":2013,"month":8,"location":"Washington D. C., USA","review":"reviewed","fields":["document-image-analysis","pattern-recognition"]},{"key":"Kunze2013_ICDARb","id":"1371","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Kai Kunze","en":""},{"ja":"Yuki Shiga","en":""},{"ja":"Shoya Ishimaru","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Reading Activity Recognition Using an Off—the—Shelf EEG— —Detecting Reading Activities and Distinguishing Genres of Documents"},"journal":{"ja":"Proc 12th International Conference on Document Analysis and Recognition (ICDAR2013)"},"pages":"96-100","year":2013,"month":8,"location":"Washington, DC, USA","review":"reviewed","presen_type":"oral","fields":[],"doi":"10.1109/ICDAR.2013.27"},{"key":"Ahmed2013_ICDAR","id":"1353","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Sheraz Ahmed","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Marcus Liwicki","en":""},{"ja":"Andreas Dengel","en":""}],"title":{"ja":"Automatic Ground Truth Generation of Camera Captured Documents Using Document Image Retrieval"},"journal":{"ja":"Proc. 12th International Conference on Document Analysis and Recognition (ICDAR 2013)"},"pages":"528-532","year":2013,"month":8,"location":"Washington, DC, USA","review":"reviewed","presen_type":"oral","fields":["document-image-analysis","pattern-recognition"],"doi":"10.1109/ICDAR.2013.111","abstract":{"ja":"In this paper a novel method for automatic ground truth generation of camera captured document images is proposed. Currently, no dataset is available for camera captured documents. It is very difficult to build these datasets manually, as it is very laborious and costly. The proposed method is fully automatic, allowing building the very large scale (i.e., millions of images) labeled camera captured documents dataset, without any human intervention. Evaluation of samples generated by the proposed approach shows that 99.98% of the images are correctly labeled. Novelty of the proposed approach lies in the use of document image retrieval for automatic labeling, especially for camera captured documents, which contain different distortions specific to camera, e.g., blur, occlusion, perspective distortion, etc."}},{"key":"Gao2013","id":"1354","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Hongxing Gao","en":""},{"ja":"Marcal Rusinol","en":""},{"ja":"Dimosthenis Karatzas","en":""},{"ja":"Josep Llados","en":""},{"ja":"Tomokazu Sato","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Key-region Detection for Document Images ---Application to Administrative Document Retrieval"},"journal":{"ja":"Proc. 12th International Conference on Document Analysis and Recognition (ICDAR 2013)"},"pages":"230-234","year":2013,"month":8,"location":"Washington, DC, USA","review":"reviewed","presen_type":"poster","fields":["document-image-analysis","pattern-recognition"],"doi":"10.1109/ICDAR.2013.53","abstract":{"ja":"In this paper we argue that a key-region detector designed to take into account the special characteristics of document images can result in the detection of less and more meaningful key-regions. We propose a fast key-region detector able to capture aspects of the structural information of the document, and demonstrate its efficiency by comparing against standard detectors in an administrative document retrieval scenario. We show that using the proposed detector results to a smaller number of detected key-regions and higher performance without any drop in speed compared to standard state of the art detectors."}},{"key":"Iwamura2013_ICDAR","id":"1344","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Masaki Tsukada","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Automatic Labeling for Scene Text Database"},"journal":{"ja":"Proc. 12th International Conference on Document Analysis and Recognition (ICDAR 2013)"},"pages":"1397-1401","year":2013,"month":8,"location":"Washington, DC, USA","review":"reviewed","presen_type":"poster","fields":["document-image-analysis","pattern-recognition"],"doi":"10.1109/ICDAR.2013.276","abstract":{"ja":"It is thought that a large quantity of data improve\nquality of recognition. A large database, however, is not easy\nto obtain. The hardest task is labeling (also known as ground\ntruthing), which usually requires human intervention. Since\nlabeling by human is laborious and costly, labeling without human\n(automatic labeling) or minimization of human intervention\n(semi-automatic labeling) are ideal scenarios. As a step toward\nrealization of the scenarios, knowing how much an automatic\nlabeling system can perform without human intervention is\nimportant. In the current paper we propose a comprehensive\nautomatic labeling technique for a scene text database, which\nperforms segmentation and labeling for unsegmented and unlabeled\ncharacter images. To our best knowledge, this is the\nfirst method to realize the comprehensive process for automatic\nlabeling for scene text databases In experiments, we confirmed\nthat the proposed method could add new unlabeled data in\nparallel with improving recognition performance of the classifier."}},{"key":"Karatzas2013","id":"1352","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Dimosthenis Karatzas","en":""},{"ja":"Faisal Shafait","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Lluis Gomez i Bigorda","en":""},{"ja":"Sergi Robles Mestre","en":""},{"ja":"Joan Mas","en":""},{"ja":"David Fernandez Mota","en":""},{"ja":"Jon Almazan Almazan","en":""},{"ja":"Lluis Pere de las Heras","en":""}],"title":{"ja":"ICDAR 2013 Robust Reading Competition"},"journal":{"ja":"Proc. 12th International Conference on Document Analysis and Recognition (ICDAR 2013)"},"pages":"1115-1124","year":2013,"month":8,"location":"Washington, DC, USA","review":"reviewed","presen_type":"oral","fields":["document-image-analysis","pattern-recognition"],"doi":"10.1109/ICDAR.2013.221","abstract":{"ja":"This report presents the final results of the ICDAR 2013 Robust Reading Competition. The competition is structured in three Challenges addressing text extraction in different application domains, namely born-digital images, real scene images and real-scene videos. The Challenges are organised around specific tasks covering text localisation, text segmentation and word recognition. The competition took place in the first quarter of 2013, and received a total of 42 submissions over the different tasks offered. This report describes the datasets and ground truth specification, details the performance evaluation protocols used and presents the final results along with a brief summary of the participating methods."}},{"key":"Kimura2013_ICDAR","id":"1355","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Takashi Kimura","en":""},{"ja":"Rong Huang","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"The Reading-life Log --- Technologies to Recognize Texts That We Read"},"journal":{"ja":"Proc. 12th International Conference on Document Analysis and Recognition (ICDAR 2013)"},"pages":"91-95","year":2013,"month":8,"location":"Washington, DC, USA","review":"reviewed","presen_type":"oral","fields":["document-image-analysis","pattern-recognition"],"doi":"10.1109/ICDAR.2013.26","abstract":{"ja":"Reading life log is a type of techniques to automatically and unconsciously record people’s reading intentions, interests and habits. Besides, it can also serve as various assistants in our daily life. In this paper, a reading-life log system is implemented by a head-mounted and unobtrusive video camera with a high resolution and a high shutter speed. We utilize DP matching, and propose a text-based frame mosaicing method to integrate multiple frames in a clip. The developed system is tested in the various environments indoor and outdoor. The experimental results show that our system can provide reliable outputs with respect to the most correct responses. The infrequent misregistration between lines also indicates the feasibility and validity of the text-based frame mosaicing."}},{"key":"Kobayashi2013_ICDAR","id":"1345","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Takuya Kobayashi","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Takahiro Matsuda","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"An Anytime Algorithm for Camera-Based Character Recognition"},"journal":{"ja":"Proc. 12th International Conference on Document Analysis and Recognition (ICDAR 2013)"},"pages":"1172-1176","year":2013,"month":8,"location":"Washington, DC, USA","review":"reviewed","presen_type":"poster","fields":["document-image-analysis","pattern-recognition"],"doi":"10.1109/ICDAR.2013.231","abstract":{"ja":"In a scene image, some characters are difficult to recognize and some others are recognized easily. Such difficult characters usually make the processing time long while easy characters are recognized in a short time. In this paper, we propose a system which recognizes each character with a proper cost for the difficulty. Through the process, easy characters are recognized early and difficult ones are recognized late. This is a desired property of an anytime algorithm that the recognition accuracy does not decrease as the time increases. In order to realize it, we propose a method which splits the recognition process into several times and accumulates the recognition results and extracted features. We also discuss what is required to realize the anytime algorithm for the scene character recognition task. Experiments reveal that the proposed method obtains recognition results of easy characters earlier than the conventional method."}},{"key":"Kunze2013_ICDARa","id":"1363","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Kai Kunze","en":""},{"ja":"Hitoshi Kawaichi","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Kazuyo Yoshimura","en":""}],"title":{"ja":"The Wordometer - Estimating the Number of Words Read Using Document Image Retrieval and Mobile Eye Tracking"},"journal":{"ja":"Proc. 12th International Conference on Document Analysis and Recognition (ICDAR 2013)"},"pages":"25-29","year":2013,"month":8,"location":"Washington, DC, USA","review":"reviewed","presen_type":"oral_and_poster","fields":["document-image-analysis"],"doi":"10.1109/ICDAR.2013.14","awards":{"en":["Best Paper Award Winner"]},"abstract":{"ja":"We introduce the Wordometer, a novel method to\nestimate the number of words a user reads using a mobile eye\ntracker and document image retrieval. We present a reading\ndetection algorithm which works with over 91 % accuracy over\n10 test subjects using 10-fold cross validation. We implement two\nalgorithms to estimate the read words using a line break detector.\nA simple version gives an average error rate of 13,5 % for 9 users\nover 10 documents. A more sophisticated word count algorithm\nbased on support vector regression with an RBF kernel reaches\nan average error rate from only 8.2 % (6.5 % if one test subject\nwith abnormal behavior is excluded). The achieved error rates\nare comparable to pedometers that count our steps in our daily\nlife. Thus, we believe the Wordometer can be used as a step\ncounter for the information we read to make our knowledge life\nhealthier."}},{"key":"Toyama2013a","id":"1361","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Takumi Toyama","en":""},{"ja":"Wakana Suzuki","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Wearable Reading Assist System: Augmented Reality Document Combining Document Retrieval and Eye Tracking"},"journal":{"ja":"Proc. ICDAR2013"},"year":2013,"month":8,"location":"Washington, DC, USA","review":"reviewed","presen_type":"oral","fields":["document-image-analysis"],"doi":"10.1109/ICDAR.2013.15"},{"key":"Shirazi2013","id":"1369","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Alireza Sahami Shirazi","en":""},{"ja":"Neils Henze","en":""},{"ja":"Tilman Dingler","en":""},{"ja":"Kai Kunze","en":""},{"ja":"Albrecht Schmidt","en":""}],"title":{"ja":"Upright or sideways?: analysis of smartphone postures in the wild"},"journal":{"ja":"Proceedings of the 15th international conference on Human-computer interaction with mobile devices and services"},"pages":"362-371","year":2013,"month":7,"location":"Munich, Germany","review":"reviewed","presen_type":"oral","fields":[],"doi":"10.1145/2493190.2493230"},{"key":"Kise2013","id":"1356","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Riki Kudo","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Shinichiro Omachi","en":""}],"title":{"ja":"A Proposal of Writing—Life Log and Its Implementation Using s Retrival—Based Camera—Pen"},"journal":{"ja":"Proc. the 16th International Graphonomics Society Conference (IGS 2013)"},"pages":"86-89","year":2013,"month":6,"location":"Nara, Japan","review":"reviewed","fields":[],"url":"http://www.m.cs.osakafu-u.ac.jp/publication_data/1356/IGS.pdf"},{"key":"Kunze2013_CHI","id":"1373","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Kai Kunze","en":""},{"ja":"Hitoshi Kawaichi","en":""},{"ja":"Kazuyo Yoshimura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Towards inferring language expertise using eye tracking"},"journal":{"ja":"CHI'13 Extended Abstracts on Human Factors in Computing Systems"},"pages":"217-222","year":2013,"month":4,"review":"reviewed","fields":[],"doi":"10.1145/2468356.2468396"},{"key":"Utsumi2013b","id":"1341","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Yuzuko Utsumi","en":""},{"ja":"Yuya Kato","en":""},{"ja":"Kai Kunze","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Who are you? — A Wearable Face Recognition System to Support Human Memory"},"journal":{"ja":"Proceedings of the 4th Augmented Human International Conference"},"pages":"150-153","year":2013,"month":3,"location":"Stuttgart, Germany","review":"reviewed","fields":["pattern-recognition"],"doi":"10.1145/2459236.2459262"},{"key":"Kawabe2013","id":"1794","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"川邉 恵一","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"黄瀬 浩一","en":""},{"ja":"中山 利宏","en":""},{"ja":"岩村 雅一","en":""}],"title":{"ja":"改札口における特定旅客認識の基礎研究"},"journal":{"ja":"鉄道サイバネ・シンポジウム論文集"},"number":"50","num_pages":"1-4","year":2013,"month":11,"review":"not_reviewed","fields":[]},{"key":"Kamimura2013","id":"1797","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"上村将之","en":""},{"ja":"フォン ヤオカイ","en":""},{"ja":"内田誠一","en":""},{"ja":"岩村雅一","en":""},{"ja":"大町真一郎","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"Reading‐Life Logにおける追跡の利用"},"journal":{"ja":"電気関係学会九州支部連合大会講演論文集"},"year":2013,"month":9,"review":"not_reviewed","fields":[]},{"key":"Kato2013","id":"1350","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"加藤 祐也","en":""},{"ja":"内海ゆづ子","en":""},{"ja":"Kai Kunze","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"記憶補助を目的とした装着型実時間顔認識システム"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2013)"},"presen_id":"DS-21","year":2013,"month":7,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Kimura2013_SSII","id":"1349","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"木村 崇志","en":""},{"ja":"フォン ヤオカイ","en":""},{"ja":"内田 誠一","en":""},{"ja":"岩村 雅一","en":""},{"ja":"大町 真一郎","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"リーディングライフ・ログ"},"journal":{"ja":"第19回 画像センシングシンポジウム(SSII2013)","en":"19th Symposium on Sensing via Image Information（SSII2013）"},"volume":"19","presen_id":"IS3-33","year":2013,"month":6,"location":"神奈川県横浜市","review":"not_reviewed","presen_type":"poster","fields":["document-image-analysis"]},{"key":"Matsuda2013_SSII","id":"1348","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"松田 崇宏","en":""},{"ja":"小林 拓也","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"リファレンスポイントを用いた情景内文字認識の高速化"},"journal":{"ja":"第19回 画像センシングシンポジウム(SSII2013)","en":"19th Symposium on Sensing via Image Information（SSII2013）"},"volume":"19","presen_id":"IS1-28","year":2013,"month":6,"location":"神奈川県横浜市","review":"not_reviewed","presen_type":"oral","fields":["pattern-recognition"]},{"key":"Utsumi2013a","id":"1324","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"内海 ゆづ子","en":""},{"ja":"坂野 悠司","en":""},{"ja":"前川 敬介","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"局所特徴量と近似最近傍探索を用いた大規模データベースに対する高速顔認識"},"book_title":"情報処理学会研究報告コンピュータビジョンとイメージメディア（CVIM）","number":"4","presen_id":"2013-CVIM-186","pages":"1-7","year":2013,"month":3,"fields":["pattern-recognition","nearest-neighbor-search"],"abstract":{"ja":"本稿では，犯罪捜査支援を目的とした大規模データベースに対する高速な候補の絞り込み手法を提案する．犯罪捜査支援を目的とした顔認識に必要とされる条件は，高速かつ高精度な絞り込みが可能であり，低解像度画像であっても，精度を落とすことなく絞り込み可能であることである．そこで，局所特徴量と近似最近傍探索に基づく特徴量の投票処理を組み合わせた手法を用いることで，高速で高精度な顔認識を実現する．また，低解像度画像に対する認識精度向上のため，Dense sampling を用いた特徴抽出を行う．1000 万顔画像データベースを作成し，顔認識実験を行った結果，精度98.5 [%] で185 [ms] の探索が\n可能であった．また，低解像度画像に対してDense sampling を用いて特徴抽出した場合，10 万枚顔画像データベースに対して精度97.7[%] で53[ms] の検索が可能となり，提案手法が低解像度画像に対しても有効に働くことが示された．"}},{"key":"Ahmed2013_IEICEPRMU","id":"1327","type":"domestic_conf","language":"EN","show":"show","authors":[{"ja":"Sheraz Ahmed","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Marcus Liwicki","en":""},{"ja":"Andreas Dengel","en":""}],"title":{"ja":"Automatic Word Ground Truth Generation for Camera Captured Documents"},"journal":{"ja":"IEICE Technical Report"},"volume":"112","number":"495","presen_id":"PRMU2012-204","pages":"141-146","year":2013,"month":3,"location":"東京都調布市","review":"not_reviewed","presen_type":"oral","fields":["document-image-analysis"]},{"key":"Kimura2013_PRMU","id":"1336","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"木村 崇志","en":""},{"ja":"フォン ヤオカイ","en":""},{"ja":"内田 誠一","en":""},{"ja":"岩村 雅一","en":""},{"ja":"大町 真一郎","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"Reading-Life Logの実装"},"journal":{"ja":"電子情報通信学会技術報告"},"volume":"112","number":"495","presen_id":"PRMU2012-219","pages":"231-236","year":2013,"month":3,"location":"東京都調布市","review":"not_reviewed","presen_type":"oral","fields":[]},{"key":"Kawashi2013","id":"1325","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"川市 仁史","en":""},{"ja":"Kai Kunze","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"文書画像検索を用いたリーディングライフログの提案"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"112","number":"495","presen_id":"PRMU2012-223","pages":"255-260","year":2013,"month":3,"location":"東京都調布市","review":"not_reviewed","presen_type":"oral","fields":["document-image-analysis"],"abstract":{"ja":"リーディングライフログとは，文書に対する「読む」という行動をデジタルデータとして記録することであり，これを解析することでユーザの趣味，嗜好に合わせた情報の提供が可能になると考えられる．本研究ではアイトラッカと文書画像検索を組み合わせることで読んでいる文書と読んでいる位置を特定し，\n質の高いリーディングライフログの実現を目指す．本稿では，その第一歩として，視点情報における改行の判定法を提案する．また，改行判定を用いて読んだ単語数を推定する手法を提案する．実験の結果，改行判定の誤差は2.1行程度であった．また，単語数の推定誤差は，単純な推定法でも13.5%，SVRを用いた推定法で8.2%となり，提案手法の有効性が示された．"}},{"key":"Kazuyo2013_PRMU","id":"1328","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"吉村 和代","en":""},{"ja":"川市 仁史","en":""},{"ja":"Kai Kunze","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"アイトラッカで取得した視点情報と文書理解度の関係"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"112","number":"495","presen_id":"PRMU2012-224","pages":"261-266","year":2013,"month":3,"location":"東京都調布市","review":"not_reviewed","presen_type":"oral","fields":["document-image-analysis"],"abstract":{"ja":"本稿では，アイトラッカで取得した視点情報を用いて，文書に対する理解度を推定する手法を提案する．\n本手法ではまず，アイトラッカで得られた視点情報に対し，眼球運動の性質を利用した特徴量を抽出し，8 次元の特徴ベクトルを生成する．そして，得られた特徴ベクトルを用いて，読み判定をし，ユーザが読んでいるか読んでいないかを判別する．さらに，理解度推定をすることで，文書に対する理解度を推測する．被験者10 人，10 文書を用いた実験の結果，読み判定については識別率91%を得，有効性を確認した．一方，文書理解度推定については，5 段階で表現される問題の正答率により理解度を表現したところ，平均二乗誤差1.059 で正答率を推定するに留まり，十分\nな精度を得るに至らなかった．本稿ではその理由についても考察する．"}},{"key":"Kudo2013_PRMU","id":"1307","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"工藤 力","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"カメラペンシステムにおけるup・down判定と筆跡位置修正の有効性評価"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"112","number":"495","presen_id":"PRMU2012-206","pages":"151-156","year":2013,"month":3,"location":"東京都調布市","review":"not_reviewed","presen_type":"oral","fields":[],"abstract":{"ja":"カメラペンシステムとは，小型カメラを取り付けたペンを使用して印刷文書への筆跡をデジタルデータと\nして復元するシステムである．このシステムは，筆跡を求める処理と，その筆跡の文書上の位置を求める処理から成\nり立っており，前者には筆跡形状復元法，後者は文書画像検索法を用いる．文書画像検索を用いて，撮影画像からデー\nタベースに登録した文書を検索し，筆跡形状復元法を用いて求めた筆跡を文書上に記録することができる．本稿では，\nこのカメラペンシステムに対し，up・down 判定法と筆跡位置修正を行う手法を加えることを提案する．up・down 判定法を加えることにより，多画数の手書きに対して筆跡を復元することができる．また，筆跡位置修正を行うことにより，ペン先の開始点がずれることから生じる問題を解消することができる．これらの手法の有効性を調べるために，\n多画数の筆記を行い，筆跡位置修正の有無による誤差の違いを調べる比較実験を行った．比較実験の結果から，これ\nらの手法は，カメラペンシステムに対して有効であることがわかった．"}},{"key":"Sugimoto2013_TR","id":"1338","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"杉本 恭隆","en":""},{"ja":"岩田 基","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"画像検索とのハイブリッド方式による文書画像検索の性能向上"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"112","number":"495","pages":"43-48","year":2013,"month":3,"location":"東京都調布市","review":"not_reviewed","presen_type":"oral","fields":[]},{"key":"Suzuki2013","id":"1326","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"鈴木 若菜","en":""},{"ja":"外山 託海","en":""},{"ja":"黄瀬 浩一","en":""},{"ja":"Andreas Dengel","en":""}],"title":{"ja":"文書に対する視点情報を利用した情報表示システムの提案"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"112","number":"495","presen_id":"PRMU2012-226","pages":"273-278","year":2013,"month":3,"location":"東京都調布市","review":"not_reviewed","presen_type":"oral","fields":["document-image-analysis"]},{"key":"Tsukada2013_PRMU","id":"1315","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"塚田 真規","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"自動ラベル付けによる情景文字画像データベース"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"112","number":"495","presen_id":"PRMU2012-220","pages":"237-242","year":2013,"month":3,"location":"東京都調布市","review":"not_reviewed","presen_type":"oral","fields":["pattern-recognition"],"abstract":{"ja":"多様な情景文字画像の認識には，多様なデータが登録された大規模文字データベースが必要となる．\n大規模データベースの構築には収集された文字画像に対して，文字領域の切り出し，データへのラベル付け，データベースへの登録という人の介在が必要な処理があり，膨大なコストがかかる．本稿では事例ベース文字検出手法とSemi-supervised learningに基づく自動ラベル付け手法の2つの手法を提案し，データベース構築に必要な処理を自動化することで構築に必要なコストを削減する．実験では各手法に対する評価実験と，2つの提案手法を組み合わせて文字領域の切り出しからデータベースへの登録までを自動化した統合システムについても評価した．"}},{"key":"Sakata2013_PRMU","id":"1323","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"阪田 智大","en":""},{"ja":"的崎 伸彰","en":""},{"ja":"黄瀬 浩一","en":""},{"ja":"岩村 雅一","en":""}],"title":{"ja":"局所特徴量の大規模照合に基づく物体インスタンスの検索　〜 TRECVID2012 Instance Searchタスク参加報告 〜"},"book_title":"電子情報通信学会技術研究報告","volume":"112","number":"441","presen_id":"PRMU2012-174","pages":"209-214","year":2013,"month":2,"review":"not_reviewed","fields":["object-recognition"],"abstract":{"ja":"我々がTRECVID 2012のInstance Searchタスクに参加した結果を報告する．\n我々の研究では，BoFのような手法を用いず局所特徴量を直接マッチングすることで，\n画像中の物体位置や背景の変化に対応できるという局所特徴量の利点を残し，それにより高い認識率を目指す．\nこの際，データ量が膨大なためマッチングに時間がかかるという問題があるが，ハッシュに基づく近似最近傍探索手法を利用することにより，この問題に対処する．\n本手法を利用し，最大で18.2%のMean Average Precisionとなり，全24チーム中4位という結果を得ることができた．\n本稿では，このような単純な手法で何がどこまで認識できるのかを，実験データに基づいて明らかにするとともに，限界とそれを克服するための方策についても考察する．"}},{"key":"Sato2013_IEICEPRMU","id":"1335","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"佐藤 智一","en":"Tomokazu Sato"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"空間インデクシングに基づく距離推定を用いた高速かつ省メモリな近似最近傍探索","en":"Fast and Memory Efficient Approximate Nearest Neighbor Search with Distance Estimation Based on Space Indexing"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE"},"volume":"112","number":"441","presen_id":"PRMU2012-142","pages":"73-78","year":2013,"month":2,"location":"大阪府堺市","review":"not_reviewed","presen_type":"oral","fields":["nearest-neighbor-search"],"abstract":{"ja":"本稿では，高次元かつ大規模なデータセットを高速かつ省メモリで扱うための近似最近傍探索手法を提案する．近似最近傍探索は，入力ベクトルに最も距離が近いベクトルを探索する最近傍探索問題において，探索誤りを許容することで計算時間，メモリ使用量を大幅に削減するものである．ハッシュ構造を用いてクエリからの距離を推定する手法に改良を加え，より一般的なデータに対して，高速に解を得られるようになった．また，ベクトルを量子化表現することでメモリ使用量を削減した．実験では大規模なSIFT特徴とGIST特徴のデータセットを用いて精度，処理時間，メモリ使用量の観点で比較を行い，最近の既存手法と比べて高速に最近傍点を得られることを確認した．"}},{"key":"Takeda2013_PRMU","id":"1346","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"竹田 一貴","en":"Kazutaka Takeda"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"1億ページのデータベースを対象とした大規模文書画像検索"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"112","number":"441","presen_id":"PRMU2012-161","pages":"131-136","year":2013,"month":2,"location":"大阪府堺市","review":"not_reviewed","presen_type":"oral","fields":["document-image-analysis","document-image-retrieval"],"abstract":{"ja":"本稿では，1億ページのデータベースを対象とした大規模実時間文書画像検索法を提案する．我々はすでに，1,000万ページのデータベースから実時間で検索可能な手法を提案している．この手法を用いてさらなる大規模化を実現するためには，より一層のメモリ削減が求められる．同時に，メモリ削減による検索精度の低下を抑制する必要がある．これを実現するため，検索に有効な特徴量のサンプリング法と，特徴量の柔軟な照合を実現するための多重探索法を提案する．1,000万ページのデータベースを用いた実験から，従来手法と比較して，検索精度を維持したまま70%のメモリ削減を実現できることが分かった．また，1億ページのデータベースから，必要メモリ量236GB，精度98.7%，処理時間26.8msで検索でき，高いスケーラビリティを持つこと確認した．"}},{"key":"Nakatsuji2013_Symp","id":"1305","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"仲辻 友博","en":"Tomohiro NAKATSUJI"},{"ja":"岩田 基","en":"Motoi IWATA"},{"ja":"黄瀬 浩一","en":"Kouichi KISE"}],"title":{"ja":"モニタに映し出された動画像から透かしを抽出できる電子透かし法","en":"Digital Watermarking Method for the Ability to Extract aWatermark from the Watermarked Video Projected on the Monitor"},"journal":{"ja":"2013年 暗号と情報セキュリティシンポジウム概要集","en":"Abstracts of the 2013 Symposium on Cryptography and information Security"},"pages":"84","year":2013,"month":1,"review":"not_reviewed","fields":[],"abstract":{"ja":"本稿では, スマートフォンを利用したサービスへ電子透かしを応用することを考える. 例え\nば, 透かし入り動画像に対してスマートフォンをかざすことによって, 透かしを抽出して情報を得るサー\nビスが考えられる. スマートフォンで透かし入り動画像を撮影すると, 撮影した動画像が傾いていたり,\n透かし入り動画像以外の部分を含んで録画される. また, 撮影した動画像はフレームレートが異なり, 色\n合いが変化していることが想定される. 本研究では撮影した動画像から透かし入り動画像の部分を検出\nし, 補正したものから透かしを抽出することを考える. また, 動画像の途中から撮影した場合でも情報を\n得られることが必要である. 動画像の途中からでも透かしを取り出せる手法として, 2007 年に山本らが時\n間同期外しに耐性をもつ電子透かし法を提案している. 山本らは動画像をブロック分割して3 次元DFT\nを施し, 一つの周波数平面のみを用いて全てのブロックに同じ3 ビットの情報を埋め込んでいた. 提案手\n法では, 複数の周波数平面を用い, 全てのブロックに個別の情報を埋め込んでも情報を抽出できるように\n変更する. 山本らは一つのフレームセットから全てのビットを抽出していたが, ブロックごとに埋め込む\nビット0, 1 に対応する系列との相関値の差の大きさが最大になる所を動画像全体を用いて探索するよう\nに抽出法を改良した. この抽出法を用い, 実際にスマートフォンで撮影した透かし入り動画像を補正した\n動画像に対して抽出実験を行い, その有用性を示す."}},{"key":"Goushi2013","id":"1427","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"合志 清一","en":""},{"ja":"栗林 稔","en":""},{"ja":"越前 功","en":""},{"ja":"岩田 基","en":""},{"ja":"川村 正樹","en":""},{"ja":"岩村 惠市","en":""}],"title":{"ja":"画像・映像信号用電子透かし評価基準2013の提案"},"journal":{"ja":"2013年 暗号と情報セキュリティシンポジウム論文集"},"pages":"3F1-1","year":2013,"month":1,"review":"not_reviewed","fields":[]},{"key":"Yamanaka2013_Symp","id":"1304","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"山中 賢次","en":"Kenji Yamanaka"},{"ja":"岩田 基","en":"Motoi Iwata"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"印刷物を対象とした電子透かし法の検討","en":"A consideration of Digital Watermarking Method to Printed Images"},"journal":{"ja":"2013年暗号と情報セキュリティシンポジウム概要集","en":"Abstracts of the 2013 Symposium on Cryptography and Information  Security"},"pages":"96","year":2013,"month":1,"review":"not_reviewed","fields":[],"abstract":{"ja":"電子透かしを用いて，印刷物に人間には認識できないように情報を埋め込み、それをスマートフォンで撮影して情報を取り出す手法について検討する．透かし入りの印刷物にユーザーがスマートフォンをかざすことで情報を得られることを目指している．これを実現するには，スマートフォンをかざしている間，撮影と，情報の抽出処理を並行して行う必要がある．そのため，従来の電子透かしでは重視されていなかった抽出処理の高速性が求められる．ほかの要件としては，撮影映像の解像度が高いと抽出処理に時間がかかるため，解像度が低い撮影映像から情報を抽出できなければならない．また，印刷の際には色合いの変化が生じたり，撮影の際には手ブレや位置ずれなどが生じたりするため，これらの外乱に耐性を持たせなければならない．"}},{"key":"Imura2013","id":"1303","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"伊村 政志","en":""},{"ja":"岩村 雅一","en":""},{"ja":"日浦 慎作","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"ピントのずれに頑健な文字認識手法の高速化"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"112","number":"385","presen_id":"PRMU2012-111","pages":"257−262","year":2013,"month":1,"fields":["pattern-recognition","computational-photography"],"abstract":{"ja":"文字認識技術には，ピントのずれによる認識率低下が問題として存在する．この問題の原因は，文字の特徴が潰れ，パターンが類似することにある．そのため本研究では，符号化開口を用いることでピントがずれても文字 がその特徴を残すようにし，ピントが合わない場合の文字認識を可能にする．前報では，予めテンプレートのぼけたパターンを用意して，認識対象とテンプレートマッチングにより認識する手法を提案した．\nしかし，この手法では処理時間が膨大なものとなってしまう．\nそこで本稿ではぼけの度合推定と文字認識の2つのプロセスに分割することで上記手法の高速化を図る．"}},{"key":"Sun2013_PhD","id":"1311","type":"doctoral_thesis","language":"EN","show":"show","authors":[{"ja":"Weihan Sun","en":""}],"title":{"ja":"Study on Content-Based Copy Detection for Copyright Protection of Line Drawings"},"year":2013,"month":3,"review":"not_reviewed","fields":["document-image-analysis","information-retrieval","pattern-recognition"]},{"key":"Katte2013","id":"1333","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"勝手 美紗","en":""}],"title":{"ja":"背景と物体の特徴量に基づくイベント検出"},"year":2013,"month":3,"review":"not_reviewed","fields":[]},{"key":"Kobayashi2013_MThesis","id":"1314","type":"master_thesis","language":"EN","show":"show_except_datafile","authors":[{"ja":"小林 拓也","en":"Takuya Kobayashi"}],"title":{"ja":"Anytime Algorithm によるカメラベース文字認識手法の高速化","en":"On An Anytime Algorithm for Camera-Based Character Recognition"},"year":2013,"month":3,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Nakatsuji2013_MThesis","id":"1316","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"仲辻 友博","en":""}],"title":{"ja":"3次元DFT係数を用いてモニタに映った動画像から情報を取得できる電子透かし法"},"year":2013,"month":3,"review":"not_reviewed","fields":[]},{"key":"Sakata2013_MThesis","id":"1313","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"阪田 智大","en":"Tomohiro Sakata"}],"title":{"ja":"近似最近傍探索を用いた大規模映像中の特定物体認識","en":"Specific Object Recognition for Large-Scale Videos with Approximate Nearest Neighbor Search"},"year":2013,"month":3,"review":"not_reviewed","fields":["object-recognition"]},{"key":"Sato2013_MThesis","id":"1334","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"佐藤 智一","en":""}],"title":{"ja":"近似最近傍探索の高速化と省メモリ化に関する研究","en":"Research on Fast and Memory—Efficient Approximate Nearest Neighbor Search"},"year":2013,"month":3,"review":"not_reviewed","fields":[]},{"key":"Takeda2013_MThesis","id":"1332","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"竹田 一貴","en":""}],"title":{"ja":"大規模文書画像検索の実現とその応用に関する研究","en":"Document Image Retrieval for a Large—scale Database and Its Application"},"year":2013,"month":3,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"]},{"key":"Tsukada2013_MThesis","id":"1312","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"塚田 真規","en":""}],"title":{"ja":"情景文字画像に対する自動ラベル付けによる大規模データベースの構築","en":"Construction of Large Database by Automatic Labeling for Scene Characters"},"year":2013,"month":3,"review":"not_reviewed","fields":[]},{"key":"Yamanaka2013_MThesis","id":"1317","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"山中 賢次","en":""}],"title":{"ja":"スマートフォンのカメラで印刷物から透かしを読み取り可能な電子透かし法"},"year":2013,"month":3,"review":"not_reviewed","fields":[]},{"key":"Kazuyo2013_BThesis","id":"1329","type":"bachelors_thesis","language":"JA","show":"show","authors":[{"ja":"吉村 和代","en":""}],"title":{"ja":"読者の視点情報に基づく文書理解度の推定"},"year":2013,"month":3,"review":"not_reviewed","fields":[]},{"key":"Konishi2013","id":"1320","type":"bachelors_thesis","language":"JA","show":"show_except_abstract","authors":[{"ja":"小西 將貴","en":"Masataka Konishi"}],"title":{"ja":"相互部分空間法の高速化"},"year":2013,"month":3,"review":"not_reviewed","fields":["pattern-recognition","object-recognition"]},{"key":"Kudo2013_BThesis","id":"1322","type":"bachelors_thesis","language":"JA","show":"show","authors":[{"ja":"工藤 力","en":""}],"title":{"ja":"up・down判定付きカメラペンシステム"},"year":2013,"month":3,"review":"not_reviewed","fields":[]},{"key":"Matsuda2013_BThesis_Reductiona","id":"1318","type":"bachelors_thesis","language":"JA","show":"show","authors":[{"ja":"松田 崇宏","en":"Takahio Matsuda"}],"title":{"ja":"リファレンスポイントを用いた情景内文字認識の高速化","en":"Reduction of Computational Time in Scene Character Recognition with Reference Point"},"year":2013,"month":3,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Sakano2013","id":"1321","type":"bachelors_thesis","language":"JA","show":"show","authors":[{"ja":"坂野 悠司","en":""}],"title":{"ja":"1000万データベースに対する高速な顔認識"},"year":2013,"month":3,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Sugimoto2013_BThesis","id":"1337","type":"bachelors_thesis","language":"JA","show":"show","authors":[{"ja":"杉本 恭隆","en":""}],"title":{"ja":"画像検索とのハイブリッド方式による文書画像検索の性能向上"},"year":2013,"month":3,"review":"not_reviewed","fields":[]},{"key":"Kunze2013_IEEE","id":"1362","type":"survey","language":"EN","show":"show","authors":[{"ja":"Kai Kunze","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Shinichiro Omachi","en":""}],"title":{"ja":"Activity Recognition for the Mind: Toward a Cognitive \"Quantified Self\""},"journal":{"ja":"IEEE Computer"},"volume":"46","number":"10","pages":"105-108","year":2013,"month":10,"review":"not_reviewed","fields":[],"doi":"10.1109/MC.2013.339"},{"key":"Iwamura2013_TELECOM","id":"1351","type":"survey","language":"JA","show":"show","authors":[{"ja":"岩村雅一","en":"Masakazu Iwamura"}],"title":{"ja":"高速・高精度な近似最近傍探索の実現","en":"Realization of Fast and Accurate Approximate Nearest Neighbor Search"},"journal":{"ja":"TELECOM FRONTIER","en":"TELECOM FRONTIER"},"volume":"80","year":2013,"month":8,"review":"not_reviewed","fields":[],"url":"http://www.scat.or.jp/frontier/frontier80/mokuji_80.html","abstract":{"ja":"コンピュータや携帯型デバイスの普及と性能向上、インターネットの普及などにより、我々が利用できるテキスト、画像、音楽、動画などのデータは日々増加している。これらの情報は、うまく活用できれば我々の生活を豊かにすると考えられるが、そのためには膨大な情報の中から所望の情報を効率よく発見できる技術が必要不可欠である。\n本稿では、情報処理において基本的な処理である「最も似ているデータをみつける」最近傍探索と呼ばれる技術に注目する。最近傍探索に近似を導入した近似最近傍探索は、大規模データに対する探索精度と計算時間において望ましい性質を持っているため、大量のデータを活用する方法として近年よく用いられている。これらの最近の手法を紹介した後、我々が提案している現在世界で最も高速な近似最近傍探索手法を紹介する。"}},{"key":"Iwamura2013_JO","id":"1343","type":"survey","language":"JA","show":"show_except_datafile","authors":[{"ja":"岩村雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬浩一","en":"Koichi Kise"}],"title":{"ja":"局所特徴量と事例検索に基づく高速特定物体認識","en":"Fast Specific Object Recognition Based on Local Feature and Example Search"},"journal":{"ja":"光学","en":"Japanese Journal of Optics"},"volume":"42","number":"8","pages":"407-412","year":2013,"month":8,"review":"not_reviewed","fields":["object-recognition"],"abstract":{"ja":"物体認識とは，画像内にある物体を認識する処理である．その中でも特に個別物体（特\n定のインスタンス）の認識を目的としたものを「特定物体認識」という．本稿では，特定\n物体認識とその応用，ならびに特定物体認識を実現する最も単純な実現方法として，局所\n特徴量と効率的な事例検索を用いる方法を紹介する．\n3","en":"Object recognition is a task to recognize an object contained in an image. If the task\nis intended to identify an instance rather than a category, it is called “specific object\nrecognition.” In the current article, we introduce the specific object recognition and its\napplications. We also introduce a simple method using local features and efficient example\nsearch to realize the specific object recognition."}},{"key":"Sun2012_IEEJ","id":"1309","type":"journal_paper","language":"EN","show":"show_except_datafile","authors":[{"ja":"Weihan Sun","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Cartoon Character Recognition Using Concentric Multi-Region Histograms of Oriented Gradients"},"journal":{"ja":"IEEJ Transactions on Electronics, Information and Systems"},"volume":"132","number":"11","pages":"1847-1854","year":2012,"month":11,"review":"reviewed","fields":["document-image-analysis","pattern-recognition","information-retrieval"],"abstract":{"ja":"Comic books are a kind of serial narrative artwork made up of comic pages. As an essential part of comics, cartoon characters appear throughout the whole series. Therefore, the recognition of cartoon characters is useful for many applications of comics. Normally, images of the same character are similar but with different representations in different scenes, such as facial expressions, poses, and viewpoints, which make them difficult to be recognized. In contrast to human being, besides face regions, there are many other parts offering the identification features for cartoon characters. In this paper, we focus on cartoon character recognition and propose Concentric Multi-Region model to explore the significant features from the parts around face regions. Histograms of Oriented Gradients (HOG) is utilized for the description of regions, and the AdaBoost algorithm is applied to obtain a new descriptor named Concentric Multi-Region Histograms of Oriented Gradients (CMR-HOG). In the experiments, 17 labeled cartoon characters are applied. Compared to other face and object recognition methods only based on face regions, the proposed method shows better performance. In addition, we proved its scalability for cartoon character recognition."}},{"key":"Klinkigt2012_IEEJ","id":"1269","type":"journal_paper","language":"EN","show":"show_except_datafile","authors":[{"ja":"","en":"Martin Klinkigt"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Object Recognition for Less-Requirements"},"journal":{"en":"IEEJ Transactions"},"volume":"131","number":"11","pages":"1878-1888","year":2012,"month":5,"review":"reviewed","fields":[]},{"key":"Takeda2012_IEICE","id":"1273","type":"letter","language":"JA","show":"show","authors":[{"ja":"竹田 一貴","en":"Kazutaka Takeda"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"大規模文書画像検索のためのメモリ効率と識別性の向上","en":"Improvements of Memory Efficiency and Discrimination Power for Real-Time Document Image Retrieval on a Large-Scale Database"},"journal":{"ja":"電子情報通信学会論文誌D","en":"Trans. IEICE"},"volume":"J95-D","number":"7","pages":"1491-1494","year":2012,"month":7,"review":"reviewed","fields":["document-image-analysis","pattern-recognition","document-image-retrieval"],"url":"https://search.ieice.org/bin/summary.php?id=j95-d_7_1491","note":{"ja":"論文のPDFファイルを<a href=\"http://www.ieice.org/jpn/trans_online/\">電子情報通信学会 Transactions online</a>からダウンロードすることができます．"},"abstract":{"ja":"大規模文書画像検索を実現するためには，必要メモリ量の削減と，特徴量の識別性向上が求められる．本稿では，特徴点のサンプリングによるメモリ削減法と，特徴量の高次元化及び冗長性の除去による安定した識別性向上法を提案する．","en":"PDF file is available at <a href=\"http://www.ieice.org/jpn/trans_online/\">IEICE Transactions online (in Japanese)</a>."}},{"key":"Sakata2012","id":"1375","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"","en":"Tomohiro Sakata"},{"ja":"","en":"Nobuaki Matozaki"},{"ja":"","en":"Koich Kise"},{"ja":"","en":"Masakazu Iwamura"}],"title":{"en":"IMP at TRECVID 2012"},"journal":{"ja":"TRECVID Workshop 2012","en":"TRECVID Workshop 2012"},"pages":"163-169","year":2012,"month":11,"location":"Gaithersburg, MD, USA","review":"reviewed","fields":["object-recognition","nearest-neighbor-search"]},{"key":"Iwata2012_ISITA","id":"1340","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"","en":"Motoi Iwata"}],"title":{"en":"Digital Watermarking Method by Selecting Suitable Pixel Value from Pixel Value Candidates Near to Original Pixel Value"},"journal":{"en":"Proc. of 2012 International Symposium on Information Theory and its Applications (ISITA2012)"},"pages":"629-633","year":2012,"month":10,"review":"reviewed","fields":[],"abstract":{"en":"In many watermarking methods, the tradeoff between\nrobustness and image quality is controlled by watermarking\nstrength. Thus the image quality was evaluated relatively\nunder the condition of keeping the robustness against assumed\nattacks. Then another axis of keeping the image quality of\nwatermarked images based on the absolute evaluation is useful\nfor the improvement of watermarking methods.\nIn this paper, we propose a new watermarking model based\non the selection of the suitable pixel value from pixel value\ncandidates near to an original pixel value. The proposed model\ncan keep the absolute image quality of watermarked images by\ndefining the set of pixel values for embedding. In the proposed\nmodel, embedding procedure is equal to the selection of the\nsuitable pixel value from the set. We use a watermarking\nmethod in color space for the implementation of the conventional\nand the proposed watermarking model because the problem of\nwatermarking methods with non-linear color transform can be\nsolved by using the proposed model."}},{"key":"Utsumi2012_ICRA","id":"1272","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Yuzuko Utsumi","en":""},{"ja":"Eric Sommerlade","en":""},{"ja":"Nicola Bellotto","en":""},{"ja":"Ian Reid","en":""}],"title":{"ja":"Cognitive Active Vision for Human Identification"},"journal":{"ja":"Proceedings of 2012 IEEE International Conference on Robotics and Automation (ICRA2012)"},"pages":"1238-1245","year":2012,"month":5,"location":"St Paul, Minnesota, USA","review":"reviewed","fields":["pattern-recognition"],"abstract":{"ja":"We describe an integrated, real-time multi-camera\nsurveillance system that is able to find and track individuals,\nacquire and archive facial image sequences, and perform face\nrecognition. The system is based around an inference engine\nthat can extract high-level information from an observed scene,\nand generate appropriate commands for a set of pan-tiltzoom\n(PTZ) cameras. The incorporation of a reliable facial\nrecognition into the high-level feedback is a main novelty of\nour work, showing how high-level understanding of a scene\ncan be used to deploy PTZ sensing resources effectively. The\nsystem comprises a distributed camera system using SQL\ntables as virtual communication channels, Situation Graph\nTrees for knowledge representation, inference and high-level\ncamera control, and a variety of visual processing algorithms\nincluding an on-line acquisition of facial images, and on-line\nrecognition of faces by comparing image sets using subspace\ndistance. We provide an extensive evaluation of this method\nusing our system for both acquisition of training data, and\nlater recognition. A set of experiments in a surveillance scenario\nshow the effectiveness of our approach and its potential for real\napplications of cognitive vision."}},{"key":"Kobayashi2012","id":"1263","type":"intl_conf","language":"JA","show":"show","authors":[{"ja":"","en":"Takuya Kobayashi"},{"ja":"","en":"Takumi Toyama"},{"ja":"","en":"Faisal Shafait"},{"ja":"","en":"Masakazu Iwamura"},{"ja":"","en":"Koichi Kise"},{"ja":"","en":"Andreas Dengel"}],"title":{"en":"Recognizing Words in Scenes with a Head-Mounted Eye-Tracker"},"journal":{"en":"Proceedings of the 10th IAPR International Workshop on Document Analysis Systems (DAS2012)"},"pages":"333-338","year":2012,"month":3,"location":"Gold Coast, Queensland, Australia","review":"reviewed","presen_type":"poster","fields":["document-image-analysis","pattern-recognition"],"doi":"10.1109/DAS.2012.74","abstract":{"en":"Recognition of scene text using a hand-held camera\nis emerging as a hot topic of research. In this paper, we\ninvestigate the use of a head-mounted eye-tracker for scene text\nrecognition. An eye-tracker detects the position of the user’s\ngaze. Using gaze information of the user, we can provide the\nuser with more information about his region/object of interest\nin a ubiquitous manner. Therefore, we can realize a service\nsuch as the user gazes at a certain word and soon obtain\nthe related information of the word by combining a word\nrecognition system with eye-tracking technology. Such a service\nis useful since the user has to do nothing but gazes at interested\nwords. With a view to realize the service, we experimentally\nevaluate the effectiveness of using the eye-tracker for word\nrecognition. The initial results show the recognition accuracy\nwas around 70% in our word recognition experiment and the\naverage computational time was less than one second per a\nquery image."}},{"key":"Sun2012_DAS","id":"1265","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Weihan Sun","en":"Weihan Sun"},{"ja":"Koichi Kise","en":"Koichi Kise"},{"ja":"Yoann Champeil","en":"Yoann Champeil"}],"title":{"ja":"Similar Fragment Retrieval of Animations by a Bag-of-features Approach","en":"Similar Fragment Retrieval of Animations by a Bag-of-features Approach"},"journal":{"ja":"Proceedings of the 10th IAPR International Workshop on Document Analysis Systems (DAS2012)","en":"Proceedings of the 10th IAPR International Workshop on Document Analysis Systems (DAS2012)"},"pages":"140-144","year":2012,"month":3,"review":"reviewed","fields":["document-image-analysis","pattern-recognition"],"abstract":{"en":"Serial animated cartoons, also called animations, are a kind of popular video documents that describe narratives by videos of cartoons. By using digital techniques, illegal users can divide animations into fragments and distribute them with-out any copyright permission. In this paper we focus on the copyright problem of animations and try to retrieve copyright infringement fragments based on key frames. Because of the huge volumes and rapid release of animations, it is impossible to store all the episodes into the database. We propose applying bag-of-features model to retrieve similar fragments based on the visual words extracted from a limited data set. In the experiments, 12 titles of animations are employed and the similar fragments outside database are applied as queries. Our method has achieved above 98% precision at 80% recall for fragments whose durations are over 140 seconds. From the results, we show that a latest release can be retrieved based on the features of the former ones from the same title."}},{"key":"Takeda2012_DAS","id":"1266","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"","en":"Kazutaka Takeda"},{"ja":"","en":"Koichi Kise"},{"ja":"","en":"Masakazu Iwamura"}],"title":{"en":"Real-Time Document Image Retrieval on a Smartphone"},"journal":{"en":"Proceedings of the 10th IAPR International Workshop on Document Analysis Systems (DAS2012)"},"pages":"225-229","year":2012,"month":3,"location":"Gold Coast, Queensland, Australia","review":"reviewed","presen_type":"poster","fields":["document-image-analysis","pattern-recognition","document-image-retrieval"],"doi":"10.1109/DAS.2012.71","abstract":{"en":"This paper presents a novel interface running on smartphones which is capable of seamlessly linking physical and digital worlds through paper documents. This interface is based on a real-time document image retrieval method called Locally Likely Arrangement Hashing. By just only pointing a smartphone to a paper document, the user can obtain its corresponding electronic document. This can easily provide the user with the information associated with the retrieved\ndocument. This relevant information can be superimposed on the display of smartphones. Therefore, we consider that with the help of this interface, the user can utilize paper documents as a new medium to display various information."}},{"key":"Tsukada2012","id":"1264","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masaki Tsukada","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Expanding Recognizable Distorted Characters Using Self-Corrective Recognition"},"journal":{"ja":"Proceedings of the 10th IAPR International Workshop on Document Analysis Systems (DAS2012)"},"pages":"327-332","year":2012,"month":3,"location":"Gold Coast, Queensland, Australia","review":"reviewed","presen_type":"poster","fields":["pattern-recognition"],"doi":"10.1109/DAS.2012.37"},{"key":"Katte2012","id":"1306","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"勝手 美紗","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"背景特徴量と物体の出現頻度に基づくイベント検出"},"book_title":"研究報告コンピュータビジョンとイメージメディア（CVIM）","number":"5","presen_id":"2012-CVIM-184","pages":"1 - 7","year":2012,"month":12,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"本稿では，インターネット上の大規模な動画像を自動で分類することを目的とし，動画像中のイベントを検出する．インターネットで配信されているホームビデオなどの動画の多くは視点が固定されずに撮影されているため，動画像から人物や物体の動きの情報を取得することは困難であり，行動認識などの技\n術を用いてイベントを検出することは困難となる．また，イベントは行われる環境や環境を構成する物体に大きく依存する．そこで，動画像中の背景や動画像中に登場する物体に着目し，イベントの検出を行う．背景からは，Opponent SIFT 特徴量をBag-of-Feature で表現したものを特徴量として抽出する．物体特徴量には，動画像から物体検出器により検出した物体の頻度と識別器の信頼度の値を用いる．それぞれ特徴量を用いて最近傍探索を行い，結果を統合することでイベントの認識を行った．評価はTRECVID2012 Multimedia Event Detection タスクのデータセットを用いて行った．その結果，特定の環境でのみ行われるイベントと動画中の物体を高い精度で検出できたイベントを検出できた．"}},{"key":"Crowley2012","id":"1372","type":"domestic_conf","language":"EN","show":"show","authors":[{"ja":"James L. Crowley","en":""},{"ja":"Kai S. Kunze","en":""},{"ja":"Paul Lukowicz","en":""},{"ja":"Albrecht Schmidt","en":""}],"title":{"ja":"Human Activity Recognition in Smart Environments"},"journal":{"ja":"Dagstuhl Reports Dagstuhl Seminar 12492"},"volume":"2","number":"12","pages":"17-36","year":2012,"month":12,"review":"not_reviewed","fields":[]},{"key":"Iiyama2012","id":"1302","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"飯山将晃","en":""},{"ja":"内海ゆづ子","en":""},{"ja":"清水彰一","en":""},{"ja":"寺林賢司","en":""},{"ja":"鳥居秋彦","en":""},{"ja":"満上育久","en":""},{"ja":"舩冨卓哉","en":""}],"title":{"ja":"MIRU2012若手プログラム実施報告","en":"Report of Young Researcher Program in MIRU 2012"},"journal":{"ja":"信学技報"},"volume":"112","number":"357","presen_id":"PRMU2012-80","pages":"61--68","year":2012,"month":12,"review":"not_reviewed","fields":[]},{"key":"Imura2012_MIRU","id":"1274","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"伊村 政志","en":""},{"ja":"岩村 雅一","en":""},{"ja":"日浦 慎作","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"ピントのずれに頑健な文字認識手法の提案"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2012)"},"presen_id":"IS1-72","pages":"1-8","year":2012,"month":8,"fields":["pattern-recognition","computational-photography"],"abstract":{"ja":"文字認識技術には，ピントのずれによる認識率低下が問題として存在する．この問題の原因は，文字の特\n徴が潰れ，パターンが類似することにある．そのため本研究では，符号化開口を用いることでピントがずれても文字\nがその特徴を残すようにし，ピントが合わない場合の文字認識を可能にする．しかし，ぼけの大きさはレンズの焦点\n面からのずれにより変化するため，認識には距離情報を既知にする必要がある．そこでパターン認識に特化した制約\nとして，画像に写っている文字がテンプレートのいずれかであるという条件の下で文字の認識を目指す．"}},{"key":"Kato2012_MIRU","id":"1275","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"加藤 祐也","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"カメラ付きヘッドマウントディスプレイを用いたウェアラブル顔認識システム"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2012)"},"year":2012,"month":8,"review":"not_reviewed","fields":[]},{"key":"Kimura2012","id":"1300","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"木村 崇志","en":""},{"ja":"柿迫 良輔","en":""},{"ja":"フォン ヤオカイ","en":""},{"ja":"内田 誠一","en":""},{"ja":"岩村 雅一","en":""},{"ja":"大町 真一郎","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"Reading-Life Logのプロトタイプ実装"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2012)"},"presen_id":"IS3-74","pages":"5","year":2012,"month":8,"location":"福岡県福岡市","review":"not_reviewed","presen_type":"oral","fields":["document-image-analysis","pattern-recognition"],"abstract":{"ja":"人間が日々目で見て読んでいる文字を自動的に認識，記録できれば，極めて有用である．我々はこれを\nReading-Life Log と呼び，人間の趣向や考えの把握，行動履歴の検索等への応用を考えている．実現のためには，視\n線情報の抽出，視線位置の文字の切り出し，文字認識技術が必要となる．本研究では，視線位追跡デバイス（アイト\nラッカ）とweb カメラを用いて視線情報を抽出し，視線位置の画像の切り出しを行う．そして，切り出した画像を\nOCR エンジンに入力し認識結果を得る．本稿では，そのプロトタイプの実装の様子と，その性能を評価するために\n行った屋内，屋外などの様々な環境や条件下での認識実験結果について論ずる．"}},{"key":"Imura2012_CVIM","id":"1270","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"伊村 政志","en":""},{"ja":"岩村 雅一","en":""},{"ja":"日浦 慎作","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"パターン認識に基づく制約を用いた符号化開口による文字認識"},"journal":{"ja":"情報処理学会研究報告コンピュータビジョンとイメージメディア(CVIM)"},"volume":"2012-CVIM-182","number":"30","presen_id":"CVIM182","pages":"1-7","year":2012,"month":5,"fields":["pattern-recognition","computational-photography"],"abstract":{"ja":"文字認識技術には，ピントのずれによる認識率低下が問題として存在する．この問題の原因は，文字の特徴が潰れパターンが類似することにある．そのため本研究では，符号化開口を用いることでピントがずれても文字がその特徴を残すようにし，ピントが合わない場合の文字認識を可能にする．しかし，ぼけの大きさはレンズの焦点面からのずれにより変化するため，認識には距離情報を既知にする必要がある．そこでパターン認識に特化した制約として，画像に写っている文字がテンプレートのいずれかであるという条件の下で距離を推定し，文字の認識を目指す．ぼけ画像はピントのあった画像とぼけ関数の畳込み積\n分で表現できる．前述の制約を用いれば，ピントのあった画像にテンプレートが存在するということになり，ぼけ画像とテンプレートの逆畳み込み積分により，距離を推定できると考えられる．しかし，実際は折り返し誤差などの信号処理問題やノイズの影響により，正しく距離を推定できない．本論文では逆畳み込みに代わる手法として，テンプレートを走査させて相関係数を求めることで距離を推定し，それを基にテンプレートをぼかすことで画像との比較から文字を認識した．"}},{"key":"Okada2012","id":"1267","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岡田 拓郎","en":""},{"ja":"柏木 隆宏","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"局所特徴量の部分空間を用いた複雑背景下での3次元特定物体認識"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"112","number":"37","presen_id":"PRMU2012-13","pages":"69-74","year":2012,"month":5,"review":"not_reviewed","fields":["pattern-recognition","object-recognition"],"abstract":{"ja":"3次元物体認識の問題の一つは，複雑な背景下での認識である．単純な背景下では高い認識率を得ることができる手法であっても，背景が複雑になると有効性が大幅に低下することもある．本稿では，我々が以前に提案した局所特徴量の部分空間を用いた認識手法を対象として，その手法が複雑背景に対してどの程度有効であるのかを実験的に検証する．認識手法は，局所特徴量の大規模な部分空間を用いて，相互部分空間法により認識するものである．背景の複雑さを変更した５種類のクエリを用いて認識実験を行い，複雑背景が与える影響について考察する．"}},{"key":"Katte2012_PRMU","id":"1247","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"勝手 美紗","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"物体と動き特徴を用いた行動認識"},"book_title":"電子情報通信学会技術研究報告","volume":"111","number":"430","presen_id":"PRMU2011-212","pages":"125 - 126","year":2012,"month":2,"review":"not_reviewed","fields":["pattern-recognition","object-recognition"],"abstract":{"ja":"高齢化が急速に進む中で，家に設置したカメラを用いて遠距離から高齢者を見守るシステムが注目を集めている．本稿では，見守りシステムでの高齢者の行動の詳細を獲得，提示することを目的とし，動画像からの人物の\n行動を検出する．人物の行動は物体などの周りの環境を伴って行われるものが大半である．そこで，物体検出と人物の動きを用いた行動認識手法を提案し，人物の動きだけでなく，人物の周りの物体を検出し，動きと周りの環境の相互の情報を用いて行動の認識を行う．人物の動きは，動画中における局所特徴量の特徴点の配置情報を特徴量として，周辺の環境は，人の周りで検出した物体の位置や大きさの比の情報を特徴量とする．認識には，動き，物体の特徴量を結合し，最近傍探索により行う．"}},{"key":"Togawa2012_PRMU","id":"1248","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"戸川 恵里","en":""},{"ja":"Martin Klinkigt","en":""},{"ja":"井上 勝文","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"自動選定された意味的属性に基づく物体認識","en":"Object Recognition Based on Automatic Selection of Semantic Attributes"},"book_title":"電子情報通信学会技術研究報告","volume":"111","number":"430","presen_id":"PRMU2011-213","pages":"127-128","year":2012,"month":2,"review":"not_reviewed","fields":["object-recognition"],"note":{"ja":"ポスター発表"},"abstract":{"ja":"本稿では，物体認識手法の中でも注目されている，意味的属性（属性）を用いる手法について述べる．本研究では，属性を用いる手法によって画像から特定物体を認識することを目的とする．しかし，属性を手作業で選定すると膨大な時間がかかるという問題がある．そこで本稿では問題解決の第一段階として，物体認識に有効な属性を自動的に得る，2 つの手法を提案する．第1 の手法は，DBpedia を用いて狭義で正確な属性を選定するものである．第2 の手法は，WordNet を用いてより意味的に広義な属性を選定するものである．これらの提案手法を用いて選定した属性で実際に特定物体認識を行い評価した．"}},{"key":"Ueda2012_PRMU","id":"1246","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"上田 敬介","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"カメラを用いたレイアウトフリー文書画像検索"},"book_title":"電子情報通信学会技術研究報告","volume":"111","number":"430","presen_id":"PRMU2011-227","pages":"185-190","year":2012,"month":2,"review":"not_reviewed","fields":["document-image-analysis"],"abstract":{"ja":"本稿では，我々が構築している，コンテンツ一致を基準としたカメラベースの文書画像検索手法における検索精度の向上法を提案する．我々はこれまでに，次のような特徴を持つ手法 (従来手法と呼ぶ) を提案した．すなわち，(1) 単語ごとの特徴抽出とクラスタリングを用いた単語画像の簡易コード化，(2) メッシュ特徴による特徴抽出，(3) 単語クラスタ ID の n-gram による索引付け，の 3 つである．しかし，従来手法には 3 つの問題点がある．単語の回転を考慮していないこと，フォントの変化に弱いこと，n-gram の特定性を考慮していないことである．これら対して，提案手法では回転処理を加えた単語の形状から特徴抽出を行い，更に，データベース文書を複数のフォントで登録するとともに，重み付けを加える．その結果，従来手法よりも精度の向上が見られた．レイアウトや撮影方法の異なる文書画像 320 枚をクエリ画像とし，データベースの画像 2,500 枚に対して検索実験を行ったところ，検索精度88.1%，処理時間 671[ms] を得た．検索精度については，従来手法の 42.8%，OCR を用いた手法の 70.3%から大幅に改善しており，処理時間についても，OCR を用いた手法の 1/5 であったことから，有効性が実証された．"}},{"key":"Uchida2012","id":"1299","type":"book","language":"EN","show":"show_except_datafile","authors":[{"ja":"Seiichi Uchida","en":""},{"ja":"Marcus Liwicki","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Data-Embedding Pen"},"book_title":"Multimedia Information Hiding Technologies and Methodologies for Controlling Data","pages":"396-411","year":2012,"month":10,"publisher":"IGI Global","review":"not_reviewed","fields":["document-image-analysis"],"doi":"10.4018/978-1-4666-2217-3.ch018"},{"key":"Iwata2012","id":"1342","type":"book","language":"EN","show":"show_except_datafile","authors":[{"ja":"","en":"Motoi Iwata"},{"ja":"","en":"Takao Ikemoto"},{"ja":"","en":"Akira Shiozaki"},{"ja":"","en":"Akio Ogihara"}],"title":{"en":"Pixel Value Adjustment for Digital Watermarking Using Uniform Color Space"},"book_title":"Watermarking Volume 1","pages":"49-66","year":2012,"month":5,"publisher":"InTech","review":"not_reviewed","fields":[]},{"key":"Iwamura2012_LNCS","id":"1271","type":"book","language":"EN","show":"show_except_datafile","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Faisal Shafait (Eds.)","en":""}],"title":{},"book_title":"Camera-Based Document Analysis and Recognition - 4th International Workshop, CBDAR 2011, Revised Selected Papers","volume":"7139","presen_id":"LNCS","year":2012,"month":3,"publisher":"Springer","review":"not_reviewed","fields":[],"doi":"10.1007/978-3-642-29364-1","url":"http://www.springer.com/computer/image+processing/book/978-3-642-29363-4","abstract":{"ja":"This book constitutes the thoroughly refereed post-workshop-proceedings of the 4th International Workshop on Camera-Based Document Analysis and Recognition, CBDAR 2011, held in Beijing, China, in September 2011. The 13 revised full papers presented were carefully selected during a second round of reviewing and improvement from numerous original submissions. Intended to give a snapshot of the state-of-the-art research in the field of camera based document analysis and recognition, the papers are organized in topical sections on text detection and recognition in scene images, camera-based systems, and datasets and evaluation."}},{"key":"Inoue2012","id":"1249","type":"doctoral_thesis","language":"JA","show":"show_except_abstract","authors":[{"ja":"井上 勝文","en":"Katsufumi Inoue"}],"title":{"ja":"局所特徴量を用いた特定物体認識に関する研究"},"year":2012,"month":2,"review":"not_reviewed","fields":["object-recognition","pattern-recognition"]},{"key":"Klinkigt2012_PhD","id":"1597","type":"doctoral_thesis","language":"EN","show":"show_except_datafile","authors":[{"ja":"Martin Klinkigt","en":""}],"title":{"ja":"Semantic Retrieval of ImagesUsing Object Recognition Systemsby Learning from Internet Communities"},"year":2012,"month":2,"review":"not_reviewed","fields":["object-recognition"]},{"key":"Asada2012","id":"1261","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"浅田 伸彦","en":""}],"title":{"ja":"厳しい射影歪みの下での文字の誤認識傾向に基づいた単語認識の精度向上"},"year":2012,"month":3,"review":"not_reviewed","fields":[]},{"key":"Chikano2012","id":"1252","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"近野 恵","en":""}],"title":{"ja":"カメラペンシステムのための筆跡復元精度の向上"},"year":2012,"month":3,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"]},{"key":"Kashiwagi2012","id":"1253","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"柏木 隆宏","en":"Takahiro KASHIWAGI"}],"title":{"ja":"大規模な部分空間を用いた3次元物体認識とその高速化","en":"3D Object Recognition Using a Large Number of Subspaces and Its Speed-Up"},"year":2012,"month":3,"review":"not_reviewed","fields":["object-recognition"]},{"key":"Maekawa2012","id":"1260","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"前川 敬介","en":"Keisuke Maekawa"}],"title":{"ja":"300万データベースに対する高速顔画像認識","en":"Fast Face Recognition on 3 million Images"},"year":2012,"month":3,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Ueda2012_MThesis","id":"1250","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"上田 敬介","en":""}],"title":{"ja":"レイアウトに依存しないカメラベース文書画像検索"},"year":2012,"month":2,"review":"not_reviewed","fields":[]},{"key":"Ikeda2012","id":"1259","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"池田 佑輝","en":""}],"title":{"ja":"大規模データベース構築の効率化を目指した連続画像中の文字領域追跡"},"year":2012,"month":3,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Imura2012_BThesis","id":"1257","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"伊村 政志","en":""}],"title":{"ja":"符号化開口によるぼけに頑健な文字認識"},"year":2012,"month":3,"fields":["pattern-recognition","computational-photography"]},{"key":"Kato2012_BThesis","id":"1262","type":"bachelors_thesis","language":"JA","show":"show","authors":[{"ja":"加藤祐也","en":""}],"title":{"ja":"顔向きにロバストな顔検出・顔認識〜ヘッドマウントディスプレイへの応用を目指して〜"},"year":2012,"month":3,"review":"not_reviewed","fields":[]},{"key":"Kawashi2012","id":"1258","type":"bachelors_thesis","language":"JA","show":"show","authors":[{"ja":"川市 仁史","en":""}],"title":{"ja":"リーディングライフログのための視点情報の分析と精度検証"},"year":2012,"month":3,"review":"not_reviewed","fields":["document-image-analysis"]},{"key":"Matozaki2012","id":"1256","type":"bachelors_thesis","language":"JA","show":"show","authors":[{"ja":"的崎 伸彰","en":""}],"title":{"ja":"局所特徴量を大規模に用いるInstance Searchの有効性"},"year":2012,"month":3,"review":"not_reviewed","fields":["object-recognition"]},{"key":"Sato2012","id":"1255","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"佐藤瞳","en":""}],"title":{"ja":"Adaboostを用いた文字検出手法の比較"},"year":2012,"month":3,"review":"not_reviewed","fields":[]},{"key":"Togawa2012_BThesis","id":"1764","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"戸川 恵里","en":""}],"title":{"ja":"自動選定された属性に基づく特定物体認識"},"year":2012,"month":3,"review":"not_reviewed","fields":[]},{"key":"Utsumi2012_J","id":"1278","type":"other","language":"JA","show":"show_except_datafile","authors":[{"ja":"内海ゆづ子","en":""},{"ja":"山本祐輔","en":""}],"title":{"ja":"学生フォーラム「物事の本質を追求し続けよう」松山隆司氏"},"journal":{"ja":"人工知能学会誌"},"volume":"24","number":"6","pages":"930-933","year":2012,"month":8,"review":"not_reviewed","fields":[]},{"key":"Iwamura2012","id":"1347","type":"other","language":"EN","show":"show_except_datafile","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Akira Horimatsu","en":""},{"ja":"Ryo Niwa","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Shinichiro Omachi","en":""}],"title":{"ja":"Affine-Invariant Character Recognition by Progressive Removing"},"journal":{"ja":"Electrical Engineering in Japan"},"volume":"180","number":"2","pages":"55-63","year":2012,"month":7,"publisher":"Wiley Periodicals, Inc.","review":"not_reviewed","fields":["pattern-recognition","document-image-analysis"],"note":{"ja":"Translated from IEEJ Transactions on Industry Applications (Denki Gakkai Ronbunshi), Vol. 131-D, No. 7, July 2011, pp. 873–879.\n\nPDF file is available at <a href=\"http://onlinelibrary.wiley.com/doi/10.1002/eej.22276/abstract\">Wiley's website</a>."},"abstract":{"ja":"Recognizing characters in scene images suffering\nfrom perspective distortion is a challenge. Although there\nare some methods to overcome this difficulty, they are\ntime-consuming. In this paper, we propose a set of affineinvariant\nfeatures and a new recognition scheme called\n“progressive removing” that can help reduce the processing\ntime. Progressive removing gradually removes less feasible\ncategories and skew angles by using multiple classifiers. We\nobserved that progressive removing and the use of the affine\ninvariant features reduced the processing time by about\n60% in comparison to a trivial algorithm without decreasing\nthe recognition rate. © 2012 Wiley Periodicals, Inc.\nElectr Eng Jpn, 180(2): 55–63, 2012; Published online in\nWiley Online Library (wileyonlinelibrary.com). DOI\n10.1002/eej.22276"}},{"key":"Klinkigt2011_IPSJ","id":"1268","type":"journal_paper","language":"EN","show":"show_except_datafile","authors":[{"ja":"","en":"Martin Klinkigt"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Using a Reference Point for Local Configuration of SIFT-like Features for Object Recognition with Serious Background Clutter"},"journal":{"en":"IPSJ Transactions on Computer Vision and Applications"},"volume":"3","pages":"110-121","year":2011,"month":12,"review":"reviewed","fields":["object-recognition"]},{"key":"Inoue2011_IEEJ","id":"1220","type":"journal_paper","language":"JA","show":"show_except_datafile","authors":[{"ja":"井上 勝文","en":"Katsufumi Inoue"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"データセットの違いが物体認識に与える影響の解析—特徴ベクトルの一致検索を用いた認識手法の場合—","en":"Analysis on the Effect of Dataset Differences for Object Recognition —For the Case of Recognition Methods Based on Exact Matching of Feature Vectors —"},"journal":{"ja":"電気学会論文誌(C)","en":"IEEJ Transactions on Electronics, Information and Systems"},"volume":"131","number":"11","pages":"1915-1924","year":2011,"month":11,"review":"reviewed","fields":["object-recognition"],"abstract":{"ja":"Specific object recognition methods based on exact matching of feature vectors are known as one of methods\nwhich can achieve high recognition performance for large-scale 3D specific object recognition. Since there\nare few common 3D object datasets whose size is sufficient to explore the effect of difference of object dataset\ncomposition and the effect of increasing number of objects for recognition, these effects for specific object\nrecognition methods based on exact matching of feature vectors are discussed insufficiently. The number of\nobjects in famous datasets (e.g., COIL-100) is around 100. Therefore, in this research, we prepare the dataset\nof 1002 3D objects by ourselves. In this paper, we will discuss the effect of dataset differences, which are\nbased on object structure, texture and the number of objects, for those methods such as the method based\non the Bloomier filter and the method based on a hash table with this dataset in addition to COIL-100."}},{"key":"Iwamura2011_IEEJ","id":"1190","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"堀松 晃","en":"Akira Horimatsu"},{"ja":"丹羽 亮","en":"Ryo Niwa"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"}],"title":{"ja":"段階的な枝刈りによるアフィン不変な文字認識","en":"Affine Invariant Character Recognition by Progressive Removing"},"journal":{"ja":"電気学会論文誌(D)","en":"IEEJ Transactions on Industry Applications"},"volume":"131","number":"7","pages":"873-879","year":2011,"month":7,"review":"reviewed","fields":["pattern-recognition","document-image-analysis"],"doi":"10.1541/ieejias.131.873","abstract":{"ja":"Recognizing characters in scene images suffering from perspective distortion is a challenge. Although there\nare some methods to overcome this difficulty, they are time-consuming. In this paper, we propose a set of\naffine invariant features and a new recognition scheme called “progressive removing” that can help reduce\nthe processing time. Progressive removing gradually removes less feasible categories and skew angles by\nusing multiple classifiers. We observed that progressive removing and the use of the affine invariant features\nreduced the processing time by about 60% in comparison to a trivial one without decreasing the recognition\nrate."}},{"key":"Kise2011","id":"1245","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Takahiro Kashiwagi","en":""}],"title":{"ja":"1.5 Million Subspaces of a Local Feature Space for 3D Object Recognition"},"journal":{"ja":"Proceedings of the 1st Asian Conference on Pattern Recognition"},"pages":"672-676","year":2011,"month":11,"review":"reviewed","fields":["object-recognition"],"awards":{"en":["Best Paper Award Winner"]},"abstract":{"ja":"We propose 3D object recognition methods \nwhose characteristic point is \nthe use of a large number of subspaces (1.5 million)\ngenerated from a billion of local features\nfor the recognition of 1002 objects.\nIn order to match query local features to a lot of \nsubspaces, a simple approximate nearest neighbor search\nis utilized. Based on this approximation\nthe proposed three methods are as follows:\na method with an ordinary subspace\nmethod (match each query local feature to subspaces),\na method of two-step matching which employs two versions\nof subspaces for efficient matching, and a method with\na mutual subspace method (both queries and models are\nrepresented as subspaces). \nThese methods are compared with two baselines\nwith and without subspaces.\nFrom experimental results,\nwe have confirmed the advantage of proposed \nmethods in both accuracy and efficiency. \nIn particular, the mutual subspace method achieves\n95\\% accuracy with processing time of 3.5 sec./query,\nwhich improves the accuracy of a baseline without subspaces about 60\\%.\nAs compared to the subspace method without approximate\nmatching, the mutual subspace method is more than 240 times\nfaster."}},{"key":"Sun2011_ICDAR","id":"1231","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Weihan Sun","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Similar Manga Retrieval Using Visual Vocabulary Based on Regions of Interest"},"journal":{"ja":"Proceedings of the 11th International Conference on Document Analysis and Recognition (ICDAR2011)","en":"Proceedings of the 11th International Conference on Document Analysis and Recognition (ICDAR2011)"},"pages":"1075-1079","year":2011,"month":10,"review":"reviewed","fields":[]},{"key":"Klinkigt2011a","id":"1234","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Martin Klinkigt","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Andreas Dengel","en":""}],"title":{"ja":"Generic and Specific Object Recognition for Semantic Retrieval of Images"},"journal":{"ja":"15th International Conference on Knowledge-Based and Intelligent Information & Engineering Systems"},"volume":"6881","pages":"397-406","year":2011,"month":9,"publisher":"Springer LNCS","review":"reviewed","fields":["object-recognition"],"abstract":{"ja":"Since the availability of large digital image collections the need for a proper management of them raises. New technologies as annotations or tagging support the user by doing this task. However, this task is time-consuming and, therefore, automatic annotation systems are requested. Working outside of controlled laboratory environments this request is challenging. In this paper we propose a system automatically adapted to the user's needs, providing useful annotations. We utilize Wikipedia to learn instances and abstract classes. With an evaluation in a complex use-case and dataset we show the possibility of such an attempt and achieve practical recognition rates of 26% on specific instance and 64% on abstract class level."}},{"key":"Klinkigt2011b","id":"1235","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Martin Klinkigt","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Heiko Maus","en":""},{"ja":"Andreas Dengel","en":""}],"title":{"ja":"Semantic Retrieval of Images by Learning from Wikipedia"},"journal":{"ja":"15th International Conference on Knowledge-Based and Intelligent Information & Engineering Systems"},"volume":"6884","pages":"212-221","year":2011,"month":9,"publisher":"Springer LNCS","review":"reviewed","fields":["object-recognition"],"abstract":{"ja":"In this paper we develop a prototype of an image management system supporting the user organizing his images with the help of semantic annotations. The system automatically contributes such annotations to close the gap between images and concepts expressing their content. We propose a novel integration into a Semantic Desktop and the usage of Wikipedia to address the common problem of initial knowledge acquisition and improvement of the performance. With an evaluation on a challenging dataset out system outperforms a purely image processing based approach by 30%."}},{"key":"Takeda2011","id":"1229","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Kazutaka Takeda","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Real-Time Document Image Retrieval for a 10 Million Pages Database with a Memory Efficient and Stability Improved LLAH"},"journal":{"ja":"Proc. 11th International Conference on Document Analysis and Recognition"},"pages":"1054-1058","year":2011,"month":9,"location":"Beijing, China","review":"reviewed","presen_type":"oral","fields":["document-image-analysis","document-image-retrieval"],"doi":"10.1109/ICDAR.2011.213","abstract":{"ja":"This paper presents a real-time document image retrieval method for a large-scale database with Locally Likely Arrangement Hashing (LLAH). In general, when a database is scaled up, a large amount of memory is required and retrieval accuracy drops due to insufficient discrimination power of features. To solve these problems, we propose three improvements: memory reduction by sampling feature points, improvement of discrimination power by increasing the number of feature dimensions and stabilizing features by reducing redundancy. From the experimental results, we have confirmed that the proposed method realizes 50% memory reduction, and achieves 99.4% accuracy and 38ms processing time for a database of 10 million pages."}},{"key":"Iwamura2011_ICDAR","id":"1226","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Takuya Kobayashi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Recognition of Multiple Characters in a Scene Image Using Arrangement of Local Features"},"journal":{"ja":"Proc. 11th International Conference on Document Analysis and Recognition (ICDAR 2011)"},"pages":"1409-1413","year":2011,"month":9,"location":"Beijing, China","review":"reviewed","presen_type":"poster","fields":["document-image-analysis","pattern-recognition"],"doi":"10.1109/ICDAR.2011.283","abstract":{"ja":"Recognizing characters in a scene helps us obtain useful information.\nFor the purpose, character recognition methods are required to\nrecognize characters of various sizes, various rotation angles\nand complex layout on complex background.\nIn this paper, we propose a character recognition method using\nlocal features having several desirable properties.\nThe novelty of the proposed method is to take into account \narrangement of local features so as to recognize multiple\ncharacters in an image unlike past methods.\nThe effectiveness and possible improvement of the method are discussed."}},{"key":"Liwicki2011","id":"1227","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Marcus Liwicki","en":""},{"ja":"Yoshida Akira","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Reliable Online Stroke Recovery from Offline Data with the Data-Embedding Pen"},"journal":{"ja":"Proc. 11th International Conference on Document Analysis and Recognition (ICDAR 2011)"},"pages":"1384-1388","year":2011,"month":9,"location":"Beijing, China","review":"reviewed","fields":["document-image-analysis","pattern-recognition"],"doi":"10.1109/ICDAR.2011.278","abstract":{"ja":"In this paper we propose a complete system for online stroke recovery from offline data. The key idea of our approach is to use a novel pen device which is able to embed meta information into the ink during writing the strokes. This pen-device overcomes the need to get access to any memory on the pen when trying to recover the information, which is especially useful in multi-writer or multi-pen scenarios. The actual data-embedding is achieved by an additional inkdot sequence along a handwritten pattern during writing. We design the ink-dot sequence in such a way that it is possible to retrieve the writing direction from a scanned image. Furthermore, we propose novel processing steps in order to retrieve the original writing direction and finally the embedded data. In our experiments we show that we can reliably recover the writing direction of various patterns. Our system is able to determine the writing direction of straight lines, simple patterns with crossings (e.g., “x” and “ll”), and even more complex patterns like handwritten words and symbols."}},{"key":"Takeda2011_CBDAR","id":"1244","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Kazutaka Takeda","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Memory Reduction for Real-Time Document Image Retrieval with a 20 Million Pages Database"},"journal":{"ja":"Proc. Fourth International Workshop on Camera-Based Document Analysis and Recognition (CBDAR2011)"},"pages":"59-64","year":2011,"month":9,"location":"Beijing, China","review":"reviewed","presen_type":"oral","fields":["document-image-analysis","document-image-retrieval"],"abstract":{"ja":"We have introduced the three improvements of Locally Likely Arrangement Hashing (LLAH) in ICDAR2011 to reduce a required amount of memory and increase discrimination power of features. In this paper, we show the experimental results which is obtained on a larger-scale database than that utilized for ICDAR2011. From experimental results, we have confirmed that the proposed method realizes 60% memory reduction and achieves 99.2% accuracy with 49ms/query processing time for the retrieval of a database of 20 million pages."}},{"key":"Sun2011_MVA","id":"1230","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Weihan Sun","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Similar Partial Copy Recognition for Line Drawings Using Concentric Multi-Region Histograms of Oriented Gradients"},"journal":{"ja":"Proceedings of the 12th IAPR Conference on Machine Vision Applications (MVA2011)","en":"Proceedings of the 12th IAPR Conference on Machine Vision Applications (MVA2011)"},"pages":"71-74","year":2011,"month":5,"review":"reviewed","fields":[]},{"key":"Kojima2011","id":"1213","type":"intl_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"Atushiro Kojima","en":""},{"ja":"Hiroshi Miki","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Object Recognition From n-grams of Human Actions without Shape Models"},"journal":{"ja":"ICIC Express Letters, Part B: Applications"},"volume":"2","number":"3","pages":"547-552","year":2011,"month":1,"review":"reviewed","fields":[]},{"key":"Maekawa2011","id":"1243","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"前川 敬介","en":""},{"ja":"内海 ゆづ子","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"100万顔画像データベースに対する34msでの照合の実現〜近似最近傍探索を用いた大規模高速顔画像検索〜"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"111","number":"353","presen_id":"PRMU2011-141","pages":"95-100","year":2011,"month":12,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"本稿では大規模データベースに対する高速顔認識手法を提案する．\n防犯や犯罪捜査を目的として顔認識を利用する場合，多数の人物を認識する必要がある．\nそのため，データベースの規模が拡大した場合に高速に認識を行うことは重要な課題である．\n従来の顔認識では数百枚程度のデータベースに対して認識性能の評価が行われているため，計算時間が考慮されていない．\n全探索などの手法は，データベース内の特徴量すべてと距離計算を行う必要があり，計算時間はデータベースの規模に比例するため，\nデータベースの規模が拡大すると認識に膨大な処理時間がかかる．\n本稿では，顔認識処理時間の削減のために近似最近傍探索を用いて特徴量の検索を行う．\nハッシュを利用し，対応点の探索範囲を限定することで，距離計算の回数を削減し処理の高速化を行う．\nWeb上から収集した100万枚の顔画像データベースに対して認識を行った結果，特徴点検出を用いる方法ではハッシュを用いない場合と比較して13,252倍高速に認識した．\nまた，Dense Samplingを用いることで最速で34 msで認識が可能となった．"}},{"key":"Ueda2011","id":"1242","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"上田 敬介","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"レイアウト変動にも対応できる文書画像検索法"},"book_title":"電子情報通信学会技術研究報告","volume":"111","number":"317","presen_id":"PRMU2011-111","pages":"25-30","year":2011,"month":11,"review":"not_reviewed","fields":["document-image-analysis"],"abstract":{"ja":"レイアウトが変更されていても，コンテンツが一致すれば検索が可能な文書画像検索法を提案する．\n文書画像検索の既存手法は大きく分けて2種類ある．\n1つは，各文字や各単語の近傍位置関係を特徴として検索する手法である．\nもう1つは，文字認識を行い，文書をコード化して検索を行う手法である．\nしかし，これらの手法にはそれぞれ問題点がある．\n前者の手法では，データベースの文書画像とレイアウトが違う文書画像を与えると検索ができなくなる．\n後者の手法では，レイアウト変動には柔軟であるが，処理時間が長いという問題点がある．\nそこで本研究では，レイアウトの変動にも対応でき，文字認識ほど厳密な処理を行わない手法を提案する．\nレイアウトの違う文書300ページをクエリ画像とし，\nデータベースの画像10,000枚に対して検索実験を行ったところ，\n認識精度93.7%，検索時間417[ms]を得た．これはOCRを用いて得られる\n認識精度99.0%には劣るものの，検索時間は1/4となっていることから\n高速な検索が可能であることがわかった．"}},{"key":"Ambai2011","id":"1233","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"安倍 満","en":"Mitsuru Ambai"},{"ja":"石川 博","en":"Hiroshi Ishikawa"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"坂上 文彦","en":"Fumihiko Sakaue"},{"ja":"佐藤 いまり","en":"Imari Sato"},{"ja":"佐藤 真一","en":"Shin'ichi Satoh"},{"ja":"杉本 茂樹","en":"Shigeki Sugimoto"},{"ja":"玉木 徹","en":"Toru Tamaki"},{"ja":"西山 正志","en":"Masashi Nishiyama"},{"ja":"阮 翔","en":"Xiang Ruan"}],"title":{"ja":"CVPR2011報告","en":"CVPR2011 Report"},"journal":{"ja":"情報処理学会研究報告","en":"IPSJ SIG Technical Report"},"volume":"2011-CVIM-179","number":"18","year":2011,"month":11,"review":"not_reviewed","fields":[],"abstract":{"ja":"2011 年6 月21 日〜23 日に米国コロラド州コロラドスプリングスで開催された国際会議CVPR2011 の概要を報告する．","en":"We give an overview of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR2011), which was held in Colorado Springs, Colorado, USA, from June 21st to 23rd, 2011."}},{"key":"Chikano2011","id":"1239","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"近野 恵","en":""},{"ja":"黄瀬 浩一","en":""},{"ja":"岩村 雅一","en":""},{"ja":"内田 誠一","en":""},{"ja":"大町 真一郎","en":""}],"title":{"ja":"カメラペンシステムにおける筆跡復元精度の向上"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"111","number":"317","presen_id":"PRMU2011-101","pages":"13-18","year":2011,"month":11,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"],"abstract":{"ja":"本稿では，我々が開発中のカメラペンシステムにおける筆跡の復元精度向上法を提案する．カメラペンシステムとは，カメラを取り付けたペンを用いて紙面に筆記した文字や図等を，筆跡そのままにデジタルデータ化する\nシステムである．このシステムは，筆跡を求める処理と，その筆跡の文書上の位置を求める処理から成り立っており，前者にはSURFトラッキング，後者は文書画像検索法を用いる．ここで，復元した筆跡にずれや欠損が含まれる等，復元精度が低いことが問題となっている．この主な原因は，SURFトラッキングの精度が低いことである．この問題を解決するために，LKトラッキングを利用する．効果を検証するため，SURFトラッキングを用いた従来手法と，LKトラッキングを用いた提案手法の比較実験を行った結果，従来手法に比べ精度の向上が見られた．"}},{"key":"Kimura2011_IEICEPRMU","id":"1237","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"木村 崇志","en":""},{"ja":"フォン ヤオカイ","en":""},{"ja":"内田 誠一","en":""},{"ja":"岩村 雅一","en":""},{"ja":"大町 真一郎","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"Reading-Lifelogの実現に向けた予備的検討"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"111","number":"317","presen_id":"PRMU2011-108","pages":"51-56","year":2011,"month":11,"location":"長崎県長崎市","review":"not_reviewed","presen_type":"oral","fields":["document-image-analysis","pattern-recognition"],"abstract":{"ja":"人間が日々目で見て読んでいる文字を自動的に認識，記録できれば，極めて有用である．\n我々はこれをReading-Lifelogと呼び，人間行動履歴の検索等への応用を考えている．\n実現のためには，視線情報の抽出，視線位置の文字の切り出し，文字認識技術が必要となる．\n本研究では，アイトラッカとwebカメラを用いて視線情報の抽出し，\n視線位置の画像の切り出しを行う．そして，切り出した画像をOCRエンジンに入力し認識結果を得る．\n本稿では，以上のシステムで屋内，屋外などの様々な環境や条件下で\n実験を行い，Reading-Lifelog実現に向けた検証を行う．"}},{"key":"Suzuki2011","id":"1236","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"鈴木 若菜","en":"Wakana Suzuki"},{"ja":"竹田 一貴","en":"Kazutaka Takeda"},{"ja":"外山 託海","en":"Takumi Toyama"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"プロジェクタを用いた情報投影による印刷文書へのインタラクティブ性の付加"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"111","number":"317","presen_id":"PRMU2011-111","pages":"69-74","year":2011,"month":11,"location":"長崎県長崎市","review":"not_reviewed","presen_type":"oral","fields":["document-image-analysis","document-image-retrieval"],"abstract":{"ja":"本稿では，印刷文書に対してプロジェクタで情報を投影し，付加するシステムを提案する．\n本システムでは，まず，Webカメラで印刷文書を撮影し文書画像を検索する．\nこれにより，情報投影の対象となる文書を特定すると共に，カメラ上での文書の位置を特定する．\n文書位置の特定により実際の印刷文書上の意図したところへ情報を投影することができる．\nまた，画像処理によってユーザの操作を認識し，それに応じた更なる情報をプロジェクタにより提示する．\nこのシステムを利用することで，電子文書，印刷文書にかかわらず同一のサービスを提供することができるようになる．"}},{"key":"Takeda2011_PRMU","id":"1240","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"竹田 一貴","en":"Kazutaka Takeda"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"スマートフォンで動作するリアルタイム文書画像検索"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"111","number":"317","presen_id":"PRMU2011-101","pages":"31-36","year":2011,"month":11,"location":"長崎県長崎市","review":"not_reviewed","presen_type":"oral","fields":["document-image-analysis","document-image-retrieval"],"abstract":{"ja":"本稿では，スマートフォンで動作する新たな文書画像検索システムを提案する．本システムは，Locally Likely Arrangement Hashing と呼ばれるリアルタイム文書画像検索法に基づいており，印刷文書を通して現実世界と仮想世界をスムーズにリンクさせるものである．スマートフォンを印刷文書にかざすだけで，対応する電子文書を検索し，それに関連付けられている情報をユーザに提示することができる．本システムによって，ユーザは印刷文書を，関連情報を取得するための新たなメディアとして利用できるようになる．"}},{"key":"Tsukada2011","id":"1238","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"塚田 真規","en":"Masaki TSUKADA"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"},{"ja":"黄瀬 浩一","en":"Koichi KISE"}],"title":{"ja":"歪んだ文字の認識と自動ラベル付け 〜大規模データベースの構築を目指して〜"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"111","number":"317","presen_id":"PRMU2011-115","pages":"93-98","year":2011,"month":11,"location":"長崎県長崎市","review":"not_reviewed","presen_type":"oral","fields":["document-image-analysis","pattern-recognition"],"note":{"ja":"PRMUの原稿に誤りがあったため、訂正版を置いてあります。修正箇所は図4(d)であり，それに伴い本文も訂正されています．"},"abstract":{"ja":"認識精度を向上させるために，膨大なデータ数を持ったデータセットは常に必要とされている．\nしかし，データ数が膨大なデータセットを作成するためには大量の画像にラベル付けする必要があり，\nコストや手間が非常にかかるといった問題が生じてしまう．\n本稿ではこの問題を解決し，膨大なデータセットを得るために，Self-corrective recognition algorithmに基づき\n自動でラベル付けを行い，学習する方法を提案する．この方法によって初めは認識できなかった歪んだ文字も\n認識することが可能となる．実験ではラベルなしデータを入力することで，どのような歪み文字が認識できたかを示し，\nまたその精度についても示す．"}},{"key":"Inoue2011_Conf","id":"1232","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"井上 勝文","en":"Katsufumi Inoue"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"登録物体数の違いがBloomier Filterを用いた特定物体認識手法に与える影響"},"journal":{"ja":"平成23年電気関係学会関西連合大会 講演論文集"},"presen_id":"P-24","pages":"618-619","year":2011,"month":10,"review":"not_reviewed","fields":["object-recognition","pattern-recognition"]},{"key":"Kimura2011","id":"1223","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"木村 崇志","en":""},{"ja":"ヤオカイ フォン","en":""},{"ja":"内田 誠一","en":""},{"ja":"岩村 雅一","en":""},{"ja":"大町 真一郎","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"視線情報を用いた文字認識"},"book_title":"電気関係学会九州支部第64回連合大会","presen_id":"08-2A-05","year":2011,"month":9,"location":"佐賀県佐賀市","review":"not_reviewed","presen_type":"oral","fields":[]},{"key":"Kashiwagi2011","id":"1222","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"柏木 隆宏","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"局所特徴量の多数の部分空間を用いた3次元物体認識とその高速化・高精度化"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"111","number":"193","presen_id":"PRMU2011-72","pages":"133-138","year":2011,"month":9,"review":"not_reviewed","fields":["object-recognition","pattern-recognition"],"abstract":{"ja":"莫大な数の部分空間を用いて3次元物体を高速に認識する手法を提案する．通常，データベース中の部分空間とクエリとなる特徴量の類似度は，その部分空間へ射影した際の射影長を計算することで求められる．\nそのため，データベース中の部分空間の数が莫大になると，全ての部分空間との類似度を計算し，物体を認識するために，莫大な処理時間が必要となる．そこで我々は、莫大な数の部分空間のデータベースに対して，近似的に探索を行い，高速に物体を認識する手法を提案する．また，近似探索を行った場合でも認識率を維持するため，近似を用いて高速に部分空間の絞り込みを行い，絞り込んだ部分空間において正確な類似度の比較を行う手法も提案する（2段階処理）．\n更に，この手法を相互部分空間法にも適用し，3次元物体の高速で高精度な認識を実現する．\n実験を行った結果，近似を用いて類似度の高い部分空間を求めて物体を認識した場合，近似を用いない場合に比べて処理時間を300分の1にすることができた．\nまた，近似を用いて相互部分空間法を行うことで，近似を用いた部分空間法と比べて，認識率が60%以上向上した．"}},{"key":"Sato2011_IEICEPRMU","id":"1224","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"佐藤 智一","en":"Tomokazu Sato"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"概算距離の精度向上による近似最近傍探索の高速化","en":"Fast Approximate Nearest Neighbor Search Based on Improved Approximate Distance"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"111","number":"193","presen_id":"PRMU2011-67","pages":"61-66","year":2011,"month":9,"location":"北海道函館市","review":"not_reviewed","presen_type":"oral","fields":["pattern-recognition"],"abstract":{"ja":"登録されたデータからクエリに最も近いものを探し出す最近傍探索問題では，\n探索誤りを許容することで計算時間を大幅に削減することができ，これを近似\n最近傍探索問題と呼ぶ．\n近似最近傍探索は一般に，\n最近傍点となる確率の高い点を選択し，\nそれらとクエリとの距離を計算するという\n2段階の処理で実現され，前者が手法の良し悪しを決定する．\n本稿では，この処理で用いる「概算距離」を計算量を増やすことなく，\nより高精度に推定することにより，\n高精度かつ高速な近似最近傍探索，を実現する手法を提案する．\n実験の結果，\n$50\\%$の精度で比較すると従来手法[1]と比べて，64次元のデータで約4倍，256次元のデータで約2.5倍の処理速度を得ることが確認できた．"}},{"key":"Iwamura2011_MIRU","id":"1221","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"武藤 大志","en":"Tomoyuki Mutoh"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"データの重複登録に基づく高速・省メモリな近似最近傍探索"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2011)"},"presen_id":"IS1-17","pages":"190-197","year":2011,"month":7,"location":"石川県金沢市","review":"not_reviewed","presen_type":"poster","fields":["nearest-neighbor-search"],"abstract":{"ja":"本稿では，近似最近傍探索において，探索精度を保ったまま\n処理時間とメモリ使用量を削減する手法を提案する．\n提案手法を使用することで，\n従来手法であるLSHの18\\%の処理時間と\n90\\%のメモリ使用量でLSHと同等の精度を実現できることを実験により確認した．\nさらに，何故このような現象が起こるのかについて，\n[1]に示されている\nLSHの探索効率の基準$\\rho$を用いて考察する．"}},{"key":"Takeda2011_MIRU","id":"1219","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"竹田 一貴","en":"Kazutaka Takeda"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"メモリ効率と安定性を向上させたLLAHによるカメラベース大規模実時間文書画像検索"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2011)"},"presen_id":"IS1-41","pages":"350-357","year":2011,"month":7,"location":"石川県金沢市","review":"not_reviewed","presen_type":"poster","fields":["document-image-analysis","document-image-retrieval"],"abstract":{"ja":"本稿では，大規模文書画像検索を実現するためのLocally Likely Arrangement Hashing (LLAH) の改善法を3つを提案する．第１は，データベースに登録する特徴点をサンプリングすることによるメモリ削減である．第2は，特徴量の次元数を増加させることによる識別性向上である．第3 は，冗長性を低下させることによる特徴量の安定性向上である．実験により，必要メモリ量が50%削減され，2000 万ページデータベースにおいて精度99.3%，処理時間54.5ms となることが確認された．"}},{"key":"Iwamura2011_IEICE","id":"1216","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"武藤 大志","en":"Tomoyuki Mutoh"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"近似最近傍探索におけるデータの重複登録","en":"Duplicate Registration of Data in Approximate Nearest Neighbor"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"year":2011,"month":6,"location":"神奈川県横浜市","review":"not_reviewed","presen_type":"oral","fields":["nearest-neighbor-search"],"abstract":{"ja":"近似最近傍探索においてデータの重複登録によって\n探索精度を保ったまま\n処理時間とメモリ使用量を削減する新たな手法を提案する．\n提案手法を使用することで，従来手法であるLSHの18\\%の処理時間と\n90\\%のメモリ使用量でLSHと同等の精度を実現できることを実験により確認した．\nさらに，この要因について\n[1]に示されている\nLSHの探索効率の基準$\\rho$を用いて考察する．","en":"We propose a new method based on duplicate registration of data\nin approximate nearest neighbor problem.\nIn the experiment, the proposed method achieved the same accuracy as\na conventional method, LSH,\nwith 18\\% of computational cost and 90\\% of memory usage needed for\nLSH.\nIn addition, we consider the phenomenon based on the\ncriterion $\\rho$ of efficiency of LSH appeared in [1]."}},{"key":"Uchida2011","id":"1215","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"Marcus Liwicki","en":"Marcus Liwicki"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"情報埋め込みペンによる手書き価値の向上","en":"The Data-Embedding Pen"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"year":2011,"month":5,"review":"not_reviewed","fields":["document-image-analysis"],"note":{"ja":"招待講演","en":"Invited Talk"}},{"key":"Asada2011","id":"1202","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"浅田 伸彦","en":"Nobuhiko Asada"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"文字誤認識の傾向を考慮したスペルチェッカーによる単語認識の精度向上","en":"Improvement of Word Recognition Accuracy with Spellchecker Based on Tendency of Recognition Error of Characters"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"presen_id":"PRMU2010-268","year":2011,"month":3,"location":"茨城県つくば市","review":"not_reviewed","presen_type":"oral","fields":["pattern-recognition"],"abstract":{"ja":"本稿では，環境中のあらゆる文字・単語を認識する「全方位認識」の実現のために，厳しい射影歪みに対\n応する手法について検討する．我々は実時間認識可能，レイアウトフリー，射影歪みに頑健といった3 つの要件を満\nたすカメラベースの単語認識手法を提案している．この手法は斜め45 度から撮影した文字を認識できる頑健さを持つ\nが，前述の全方位認識に際しては，射影歪みに対する更なる頑健性が求められる．そこで本稿では，前述の単語認識\n手法の文字認識誤りをオープンソースのスペルチェッカであるGNU Aspell を利用して補正する．その際，Aspell が\n持つ音素の類似性に基づくMetaphone という仕組みを文字誤認識傾向に基くMetashape に置き換える．提案手法を\n用いて実験を行った結果，文字が書かれた紙面に対して撮影角度が20 度のときに単語認識精度には最大で約24%か\nら約74%の向上が見られた．","en":"In this paper, we discuss a method to cope with severe perspective distortion to realize \\all-round recog-\nnition\" of characters and words in the environment. We have proposed a word recognition method satisfying three\nrequirements: ready for real-time processing, free from layout constraints and robust to perspective distortion. The\nmethod has the capability to recognize characters captured from the angle of 45 deg. However, in order to realize\nthe all-round recognition, further robustness is required. Thus in this paper, we introduce the GNU Aspell to rectify\nthe recognition results based on the tendency of recognition errors. For the sake of that, we introduce Metashape\nbased on the recognition errors instead of Metaphone of the Aspell based on phonemes. In the experiments, we\nfound that the proposed method improves the recognition rates for words captured from the angle of 20 deg. from\nabout 24% to about 74% at largest."}},{"key":"Kobayashi2011_IEICEPRMU","id":"1200","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"小林 拓也","en":"Takuya Kobayashi"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"局所特徴の位置関係を用いた情景画像中の文字認識","en":"Recognition of Characters in Scenes by Using Geometric Contexts of Local Features"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"110","number":"467","presen_id":"PRMU2010-275","pages":"223-228","year":2011,"month":3,"location":"茨城県つくば市","review":"not_reviewed","presen_type":"oral","fields":["document-image-analysis"],"abstract":{"ja":"カメラで撮影した情景画像中の文字を全て認識し，その中から重要な情報を得られれば非常に有用であると考えられる．それを実現するためには，認識対象の画像中から文字の部分を正確に識別し，認識することが必要となる．本稿では局所特徴の位置関係を用いることで画像中から文字部分を見つけ，その文字が何であるか認識することを考える．提案手法を用いて行った実験では，正面から撮影した情景画像中の日本語を84.3%の再現率と97.0%の適合率で認識することができた","en":"If you can get useful information from a character image in scene you take with your camera,\nall you have to do is release the shutter and you can save time.\nIn order to realize such a system, we have to propose a character recognition system such that\nit can correctly find the character area from the scene image and recognize it.\nIn this paper, we propose two methods to achieve it, which consider geometric contexts of local features.\nWe had a utmost 84.3% recall and 97.0% precision in character recognition experiments we had."}},{"key":"Saitou2011","id":"1798","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"齋藤宙也","en":""},{"ja":"菅谷至寛","en":""},{"ja":"大町真一郎","en":""},{"ja":"内田誠一","en":""},{"ja":"岩村雅一","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"サンプル文字画像からの文字パターンの合成","en":"Generation of Character Patterns from Sample Character Images"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"110","number":"467","presen_id":"PRMU2010-287","pages":"293-298","year":2011,"month":3,"review":"not_reviewed","fields":["document-image-analysis"]},{"key":"Takeda2011_IEICEPRMU","id":"1197","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"竹田 一貴","en":"Kazutaka Takeda"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"1,000万ページのデータベースを対象とした実時間文書画像検索のためのメモリ削減と安定性向上","en":"Memory Reduction and Stability Improvement for Real-Time Document Image Retrieval with a Database of 10 Million Pages"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"110","number":"467","presen_id":"PRMU2010-275","year":2011,"month":3,"location":"茨城県つくば市","review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"],"abstract":{"ja":"本稿では，Locally Likely Arrangement Hashing (LLAH)を用いた実時間検索が可能な大規模文書画像検索法を提案する．LLAHは高いロバスト性を実現するために，必要メモリ量が多いという問題がある．また，大規模化に対処するには，特徴量の識別性・安定性が十分でないという問題もある．これらの問題点を解決するため，以下の3点の改良を施す．第1は，ハッシュに保存する特徴点をサンプリングすることによる必要メモリ量の削減である．第2は，特徴量の次元数を増加させることによる識別性向上である．第3は，特徴量の冗長な次元を削除することによる安定性向上である．実験により，これらの改善によって必要メモリ量が約50% 削減されることが確認された．また，提案手法を用いて1,000万ページのデータベースを作成し実験を行ったところ，精度99.4% ，処理時間38msで検索可能であることが確認された．","en":"In this paper, we propose a large-scale document image retrieval method which realizes real-time processing with Locally Likely Arrangement Hashing (LLAH). Although LLAH has high accuracy and robustness, it requires a large amount of memory. It is also required to increase the discrimination power and stability of features for scaling up the database. For these purposes, we introduce the following three improvements. The first one is reduction of the required amount of memory by sampling feature points stored in the database. The second improvement is to increase the discrimination power by increasing the dimension of features. The last one is advancement of stability by removing redundant dimensions of features. From the experimental results, we have confirmed that the proposed improvements help to realize accuracy of 99.4% and processing time of 38ms for the database of 10 million pages."}},{"key":"Sato2011_RFDE","id":"1196","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"佐藤 智一","en":"Tomokazu Sato"},{"ja":"武藤 大志","en":"Tomoyuki Mutoh"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Kouichi Kise"}],"title":{"ja":"バケット距離に基づく近似最近傍探索","en":"Nearest Neighbor Search Based on Bucket Distance"},"journal":{"ja":"第3回データ工学と情報マネジメントに関するフォーラム論文集","en":"Proc. 3rd Forum on Data Engineering and Information Management"},"presen_id":"E2-6","pages":"1-6","year":2011,"month":2,"location":"静岡県伊豆市","review":"not_reviewed","fields":["nearest-neighbor-search"]},{"key":"Izuno2011","id":"1192","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"泉野 和弘","en":"Kazuhiro Izuno"}],"title":{"ja":"前置詞における英文自動添削システム","en":"Automatic Correction System of Preposition in English Text"},"year":2011,"month":3,"review":"not_reviewed","fields":[]},{"key":"Sasano2011","id":"1195","type":"master_thesis","language":"JA","show":"show_except_abstract","authors":[{"ja":"笹野 泰正","en":"Yasumasa Sasano"}],"title":{"ja":"HOG特徴のn-gram表現による人物行動に基づく物体認識","en":"Object Recognition Based on Human Activity Using n—grams of HOG features"},"year":2011,"month":3,"review":"not_reviewed","fields":["object-recognition","pattern-recognition"]},{"key":"Toyama2011","id":"1193","type":"master_thesis","language":"EN","show":"show_except_datafile","authors":[{"ja":"Takumi Toyama","en":"Takumi Toyama"}],"title":{"ja":"Gaze Guided Object Recognition Using a Head-Mounted Eye Tracker","en":"Gaze Guided Object Recognition Using a Head-Mounted Eye Tracker"},"year":2011,"month":3,"review":"not_reviewed","fields":["object-recognition"]},{"key":"Ujihara2011","id":"1194","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"氏原 慎弥","en":"Shinya UJIHARA"}],"title":{"ja":"二段階検索を用いた平面物体認識の大規模化","en":"Scaling-Up Planar Object Recognition Using Two-stage Search"},"year":2011,"month":3,"review":"not_reviewed","fields":["object-recognition"]},{"key":"Muto2011","id":"1191","type":"master_thesis","language":"JA","show":"show_except_abstract","authors":[{"ja":"武藤 大志","en":""}],"title":{"ja":"ハッシュ構造を用いた近似最近傍探索のトレードオフ解析と処理時間・メモリ使用量の削減"},"year":2011,"month":2,"fields":["nearest-neighbor-search"]},{"key":"Kobayashi2011_BThesis","id":"1201","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"小林 拓也","en":""}],"title":{"ja":"局所特徴の位置関係を用いた情景画像中の文字認識"},"year":2011,"month":3,"review":"not_reviewed","fields":["document-image-analysis"]},{"key":"Sakata2011","id":"1198","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"阪田智大","en":""}],"title":{"ja":"学習データの増加が識別方式に及ぼす影響の比較"},"year":2011,"month":3,"review":"not_reviewed","fields":["object-recognition"]},{"key":"Sato2011_BThesis","id":"1214","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"佐藤智一","en":"Tomokazu Sato"}],"title":{"ja":"バケット距離に基づく近似最近傍探索","en":"Nearest Neighbor Search Based on Bucket Distance"},"year":2011,"month":3,"review":"not_reviewed","fields":["information-retrieval"]},{"key":"Takeda2011_BThesis","id":"1203","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"竹田 一貴","en":""}],"title":{"ja":"大規模文書画像検索のためのメモリ削減と識別性向上"},"year":2011,"month":3,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"]},{"key":"Kise2011_J","id":"1225","type":"survey","language":"JA","show":"show_except_datafile","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"特徴点の高速照合とその文書画像検索への応用","en":"Fast Matching of Feature Points and Its Application to Document Image Retrieval"},"journal":{"ja":"電子情報通信学会誌","en":"The Journal of the Institute of Electronics, Information and Communication Engineers"},"volume":"94","number":"8","pages":"836-841","year":2011,"month":8,"address":"東京","review":"not_reviewed","fields":["document-image-analysis","pattern-recognition"],"abstract":{"ja":"特徴点の集合で表された対象の照合は，画像認識だけではなく，バイオインフォマティックスなど様々な課題に登場する基本的な問題である．従来，この問題は，Geometric　Hashingと呼ばれるアルゴリズムによって解決が図られてきた．しかしながら，この手法には対象が大規模化すると計算量の問題が生じるという弱点がある．これに対して筆者らは最近，LLAH(Locally Likely Arrangement Hashing)と呼ぶ新しい手法を提案し，大規模な特徴点照合（600点と60億点の照合）が単一CPUでも高速 (60ms以内) に行えることを示した．本稿ではこのアルゴリズムについて概説するとともに，文書画像検索への応用についても述べる．"}},{"key":"Inoue2010_IEICE","id":"1155","type":"journal_paper","language":"JA","show":"show_except_datafile","authors":[{"ja":"井上 勝文","en":"Katsufumi Inoue"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"特定物体認識におけるBloomier Filterを用いたメモリ削減法とその実験的評価","en":"Memory Reduction Method for Specific Object Recognition with the BloomierFilter and Its Experimental Evaluation"},"journal":{"ja":"電子情報通信学会論文誌D","en":"Trans. IEICE"},"volume":"J93-D","number":"8","pages":"1407-1416","year":2010,"month":8,"review":"reviewed","fields":["object-recognition"],"note":{"ja":"論文のPDFファイルを<a href=\"http://www.ieice.org/jpn/trans_online/\">電子情報通信学会 Transactions online</a>からダウンロードすることができます．","en":"PDF file is available at <a href=\"http://www.ieice.org/jpn/trans_online/\">IEICE Transactions online (in Japanese)</a>."},"abstract":{"ja":"局所特徴量を表す特徴ベクトルを用いた特定物体認識において，メモリ使用量を削減することは，重要な課題の1 つである．メモリ使用量を削減する一般的な手法として，“bag of features”モデルに基づくものがある．この手法では，クラスタリングにより複数の特徴ベクトルをある代表ベクトルにまとめることでメモリ使用量を削減している．しかし，得られる代表ベクトルの識別性が低下するため，高い認識率が得られないという問題がある．この問題を解決する1 つの方法は，特徴ベクトルを代表ベクトルに置き換えずにそのまま保持して認識処理に用いることである．この際，異なるメモリ削減の方法が必須となる．本論文では，特徴ベクトルの照合に距離計算ではなく，Bloomier filter を用いることでメモリ使用量を削減する手法を提案する．55 個の3次元特定物体を用いた実験より，“bag of features”モデルに基づく手法と比較し，同じ認識率を得るために必要なメモリ使用量を1/3 に削減することができた．また，1 万個の平面物体を用いた実験でも同様に，メモリ使用量を1/7 に削減することができた．","en":"Reduction of the amount of memory is an important issue for specific object recognition. The bag-of-features representation (BOF) based on vector quantization of feature vectors is known to be effective for the reduction. However, BOF also reduces the discriminative power of feature vectors, which results in lowering the recognition accuracy. To solve this problem, we propose a new memory reduction method using the Bloomier filter which is a probabilistic and compact data structure for storing a large amount of feature vectors. Experimental results show that the proposed method achieves 1/2 or more reduction as compared with conventional methods."}},{"key":"Kise2010_IEICE","id":"1151","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":"Koichi KISE"},{"ja":"野口 和人","en":"Kazuto NOGUCHI"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"}],"title":{"ja":"参照特徴ベクトルの増加による低品質画像の高速・高精度認識","en":"Robust and Efficient Recognition of Low Quality Images by Increasing Reference Feature Vectors"},"journal":{"ja":"電子情報通信学会論文誌D","en":"Trans. IEICE"},"volume":"J93-D","number":"8","pages":"1353-1363","year":2010,"month":8,"review":"reviewed","fields":["object-recognition"],"url":"http://search.ieice.org/bin/summary.php?id=j93-d_8_1353","note":{"ja":"論文のPDFファイルを<a href=\"http://www.ieice.org/jpn/trans_online/\">電子情報通信学会 Transactions online</a>からダウンロードすることができます．","en":"PDF file is available at <a href=\"http://www.ieice.org/jpn/trans_online/\">IEICE Transactions online (in Japanese)</a>."},"abstract":{"ja":"局所特徴量の照合による特定物体認識を考える．一般に，生成型学習を\n用いて，索引付けのための局所特徴量（参照特徴ベクトル）の数を増やせ\nば，それだけ認識率が向上する．本論文では，認識率の向上だけではなく，\n処理時間も短縮可能であることを示す．参照特徴ベクトルの\n数が増えると処理時間が短縮される，という逆説的な効果は，\n認識器の多段階化による認識処理の早期終了によって得られる．\nこれは，生成型学習によって照合に必要な探索空間を制限できるという効果による．\n1万画像を用いた認識実験の結果，6.6倍の参照特徴ベクトルを用いることで，\n処理時間を2/3, 認識率を12.2\\%改善できること，ならびに26億個の\n参照特徴ベクトルを用いて索引付けされた100万物体を，\n59ms/query, 90\\%の認識率で認識可能であることを示す．","en":"This paper concerns specific object recognition\nbased on local features.\nIn general, increasing the number of local features for indexing \n(reference feature vectors) by generative learning\nenables us to improve the recognition rate.\nIn this paper, we show that generative learning is also effective\nfor shorten the processing time. This paradoxical\neffect (i.e., shorter time with more reference feature \nvectors)\nis achieved by cascading recognizers. Generated\nlocal features allow us to terminate the recognition\nprocess earlier. In other words, generative learning\nis effective to reduce the search space for finding\nnearest neighbors. From the experimental results\nusing 10,000 reference images, 6.6 times reference feature\nvectors enabled us both to reduce the processing time \nto 2/3 from the original, and to improve the recognition\nrate by 12.2\\%. Another experiment with 1 million reference\nimages indexed by 2.6 billion reference feature vectors\nyielded the recognition rate of 90\\% in 59ms/query."}},{"key":"Sun2010_IEICE","id":"1153","type":"journal_paper","language":"JA","show":"show_except_datafile","authors":[{"ja":"孫 維瀚","en":"Weihan Sun"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"線画の著作権保護のための部分的複製検出法","en":"Partial Copy Detection for Copyright Protection of Line Drawings"},"journal":{"ja":"電子情報通信学会論文誌D","en":"Trans. IEICE"},"volume":"J93-D","number":"6","year":2010,"month":6,"review":"reviewed","fields":["document-image-analysis","information-retrieval"],"note":{"ja":"論文のPDFファイルを<a href=\"http://www.ieice.org/jpn/trans_online/\">電子情報通信学会 Transactions online</a>からダウンロードすることができます．","en":"PDF file is available at <a href=\"http://www.ieice.org/jpn/trans_online/\">IEICE Transactions online (in Japanese)</a>."},"abstract":{"ja":"図表や漫画などの線画は，直線また曲線で描く図形であり，画像出版物の中で重要なものの一つである．線画の著作権を保護するため，不正コピーを自動的に検出する手法が求められている．実際に，不正利用者は，線画をそのまま使うだけではなく，オリジナルの一部分，すなわち部分的複製を利用することが多い．その際には，部分的複製は他の素材の中に埋め込まれる形で使われる．加えて，部分的複製はそのままの形で埋め込まれるとは限らず，様々な処理によって改変されることも考えられる．その結果，不正利用の検出が困難となっている．更に，線画の場合，改変の手法として手書き複製が考えられるが，これが問題をより困難なものとしている．これらの問題に対処するため，本論文では，局所特徴量の照合による画像検索技術を用いて，不正コピーを検索する手法を提案する．具体的には，局所特徴領域の抽出手法としてMSER を，特徴量の記述子としてHOG を用いる．そして，データベースの大きさと検索時間を減少するため，主成分分析で特徴ベクトルを圧縮すると共に近似最近傍探索を用いる．実験により，提案手法は，印刷線画の部分的複製ばかりでなく，複雑な背景に埋め込まれた手書き線画に対しても有効性があることを示す．","en":"Line drawings are a type of images that consist of distinct straight and curved lines. Such as graphs and comics, line drawings are an important part of image publications, which have requirements\nof copyright protection. In practice, illegal users not only use the whole image directly, but also copy some\nparts from the original ones, which are called partial copies. In that case, partial copies can be embedded into other images. In addition, various kinds of image transformations can also be applied. Especially, in the case of line drawings, handwriting can be applied as a method of producing illegal copies, which are more difficult to be detected. To solve this problem, we apply image retrieval using local feature matching to detect illegal copies. Specifically, we propose to apply MSER as a region detector and HOG as a feature descriptor, and compress the dimensions of local feature vectors by using PCA to reduce the database size and detection time. From experimental results, we show the effectiveness of the proposed method for detecting not only printed partial copies but also handwritten ones from complex backgrounds."}},{"key":"Iwamura2010_IEICE_General","id":"1132","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"古谷 嘉男","en":"Yoshio Furuya"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"内田 誠一","en":"Seiichi Uchida"}],"title":{"ja":"付加情報の一般的な割り当て","en":"A General Assignment of Supplementary Information"},"journal":{"ja":"電子情報通信学会論文誌D","en":"Trans. IEICE"},"volume":"J93-D","number":"5","pages":"579-587","year":2010,"month":5,"review":"reviewed","fields":["pattern-recognition"],"url":"http://search.ieice.org/bin/summary.php?id=j93-d_5_579","note":{"ja":"論文のPDFファイルを<a href=\"http://www.ieice.org/jpn/trans_online/\">電子情報通信学会 Transactions online</a>からダウンロードすることができます．","en":"PDF file is available at <a href=\"http://www.ieice.org/jpn/trans_online/\">IEICE Transactions online (in Japanese)</a>."},"abstract":{"ja":"特徴量のみでは本質的に避けることができない誤認識を回避するために，\n付加情報を用いるパターン認識という枠組みが提案されている．\nこの方式では，パターン認識を行う際に，\n付加情報と呼ばれるクラスの決定を補助する少量の情報を特徴量と\n同時に用いて認識性能の改善を目指す．\n付加情報は自由に設定でき，通常は誤認識率が最小になるように設定する．\nここで問題となるのは，誤認識率が最小になる付加情報の設定方法である．\n常に正しい付加情報が得られるいう理想的な条件においては\n既に問題が定式化され，付加情報の割り当て方法が導かれている．\nしかし，実環境での使用を考えると，\n付加情報に生じる観測誤差を考慮した割り当て方法が求められる．\nそこで本論文では\n付加情報の観測誤差を考慮に入れて，問題を新たに定式化する．\nこれは付加情報が誤らない場合にも有効な一般的なものである．\n本論文で導いた割り当て方法が有効に機能することを\nマハラノビス距離を用いた実験で例示する．","en":"{\\it Pattern recognition with supplementary information} is a new\npattern recognition framework to avoid unavoidable errors.\nIt determines an output class by\ncombining a feature vector extracted from the pattern and\na small {\\it hint} on the true class.\nThe hint called {\\it supplementary information} is freely designed\nso as to minimize the error rate.\nUnder the assumption that supplementary information is infallible,\nan approximately best assignment of the information is derived.\nIn this paper, we derive a general assignments methodology of\nboth fallible and infallible supplementary information."}},{"key":"Iwamura2010_IEICE_Analysis","id":"1130","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"特徴点の配置に基づく画像検索手法の解析 —Geometric HashingとLLAHの比較と解析—","en":"Analysis on Image Retrieval Method Based on Arrangements of Feature Points —Comparison of Geometric Hashing and LLAH—"},"journal":{"ja":"電子情報通信学会論文誌D","en":"Trans. IEICE"},"volume":"J93-D","number":"4","pages":"494-501","year":2010,"month":4,"review":"reviewed","fields":["document-image-retrieval","document-image-analysis"],"url":"http://search.ieice.org/bin/summary.php?id=j93-d_4_494","note":{"ja":"論文のPDFファイルを<a href=\"http://www.ieice.org/jpn/trans_online/\">電子情報通信学会 Transactions online</a>からダウンロードすることができます．","en":"PDF file is available at <a href=\"http://www.ieice.org/jpn/trans_online/\">IEICE Transactions online (in Japanese)</a>."},"abstract":{"ja":"画像が複数の特徴点のみで記述されているとき，\n同一の配置を持つ画像を検索するにはどうすればよいだろうか．\n画像は射影歪みによって歪んでいるかもしれないし，\n特徴点の配置が完全に同一ではなく，\n特徴点の挿入や消失があるかもしれない．\nこの問題を解決する手法として，\nGeometric Hashing（GH）がよく知られている．\n一方，我々はLocally Likely Arrangement Hashing（LLAH）という手法を提案\nしている．\nいずれの手法も全探索より効率的に画像を検索できるが，\n計算時間とメモリ使用量においてはLLAHが優れている．\n本論文ではGHからLLAHに段階的に変化させることにより，\n効率的な探索に寄与する要因を考察する．","en":"For an image described with feature points,\nhow do we search the identical arrangement of feature points?\nThe image may be affected by perspective distortion and\nthe arrangement may be not completely identical due to replacement,\ninsert and deletion of feature points.\nAs methods to solve the problem, Geometric Hashing (GH)  and\nLocally Likely Arrangement Hashing (LLAH) are known.\nBoth of them can search efficiently, however, \nLLAH outperforms GH in computational cost and memory amount.\nIn this paper, we reveal the reason by comparing two methods."}},{"key":"Utsumi2010_Trans","id":"1276","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"内海ゆづ子","en":""},{"ja":"松本雄大","en":""},{"ja":"岩井儀雄","en":""}],"title":{"ja":"分枝限定法を用いた顔認識の効率化"},"journal":{"ja":"情報処理学会論文誌"},"volume":"51","number":"2","pages":"217-228","year":2010,"month":2,"review":"reviewed","fields":[]},{"key":"Uchida2010_IEICE","id":"1116","type":"letter","language":"JA","show":"show_except_datafile","authors":[{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"伊東 克啓","en":"Katsuhiro Ito"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"ペン先画像からの手書き復元","en":"Reconstruction of Handwritings via Pen-Tip Camera Images"},"journal":{"ja":"電子情報通信学会論文誌D","en":"Trans. IEICE"},"volume":"J93-D","number":"1","pages":"64-67","year":2010,"month":1,"review":"reviewed","fields":["document-image-analysis"],"url":"http://search.ieice.org/bin/summary.php?id=j93-d_1_64","abstract":{"ja":"本論文では，ペン先カメラの映像から，\n手書きパターンを復元できることを実証する．具体的\nには，紙面の微細構造ー紙指紋ーの動きに着目したビ\nデオモザイキング法により，ペン先の軌跡すなわち手\n書きパターンを推定できることを示す．","en":"This paper shows experimentally that a handwriting pattern on a paper sheet\ncan be reconstructed from a frame image sequence captured from\na pen-tip camera. The reconstruction is done by a video-mosaicing\ntechnique where precise structure of paper surface, called paper\nfingerprint, is fully utilized for estimating camera movement,\ni.e., pen movement."}},{"key":"Klinkigt2010_ACV","id":"1179","type":"intl_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"Martin Klinkigt","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"From Local Features to Global Shape Constraints: Heterogeneous Matching Scheme for Recognizing Objects Under Serious Background Clutter"},"journal":{"en":"Asian Conference on Computer Vision"},"volume":"4","pages":"2150-2161","year":2010,"month":11,"publisher":"Springer","review":"reviewed","fields":[],"abstract":{"ja":"Object recognition in computer vision is the task to categorize images based on their content. With the absence of background clutter in images high recognition performance can be achieved.\nIn this paper we show how the recognition performance is improved even with a high impact of background clutter and without additional information about the image.\nFor this task we segment the image into patches and learn a geometric structure of the object. In evaluations we first show that our system is of comparable performance to other state-of-the-art system and that for a difficult dataset the recognition performance is improved by 13.31%."}},{"key":"Sun2010","id":"1181","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Weihan Sun","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Similar Partial Copy Detection of Line Drawings Using a Cascade Classifi\fer and Feature Matching"},"journal":{"ja":"International Workshop on Computational Forensics"},"pages":"121-132","year":2010,"month":11,"review":"reviewed","fields":["document-image-analysis","pattern-recognition"],"abstract":{"ja":"Copyright protection of image publications is an important\ntask of forensics. In this paper, we focus on line drawings, which are represented\nby lines in monochrome. Since partial copies and similar copies\nare always applied in plagiarisms of line drawings, we propose combining\nthe technique of object detection and image retrieval to detect similar\npartial copies from suspicious images: first, detecting regions of interest\n(ROIs) by a cascade classifier; then, locate the corresponding source\nparts from copyrighted images using a feature matching method. The\nexperimental results have proved the effectiveness of proposed method\nfor detecting similar partial copies from complex backgrounds."}},{"key":"Iwata2010_ICFHR","id":"1174","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Kazumasa Iwata","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Shinichiro Omachi","en":""}],"title":{"ja":"Tracking and Retrieval of Pen Tip Positions for an Intelligent Camera Pen"},"journal":{"ja":"Proc. 12th International Workshop on Frontiers in Handwriting Recognition (ICFHR2010)"},"pages":"277-282","year":2010,"month":11,"location":"Kolkata, India","review":"reviewed","fields":["document-image-analysis","document-image-retrieval"],"doi":"10.1109/ICFHR.2010.50","awards":{"en":["Best Paper Award Winner"]},"abstract":{"ja":"This paper presents a method of recovering digital ink\nfor an intelligent camera pen, \nwhich is characterized by the functions\nthat (1) it works on ordinary paper and \n(2) if an electronic document is printed on the paper\nthe recovered digital ink is associated with the document.\nTwo technologies called paper fingerprint and document\nimage retrieval are integrated for realizing the above functions.\nThe key of the integration is the introduction of image mosaicing\nand fast retrieval of previously seen fingerprints based on\nhashing of SURF local features.\nFrom the experimental results of 50 handwritings,\nwe have confirmed that the proposed method is effective\nto recover and locate the digital ink from the handwriting\non a physical paper."}},{"key":"Liwicki2010_ICFHR","id":"1175","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Marcus Liwicki","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Embedding Meta-Information in Handwriting —Reed-Solomon for Reliable Error Correction"},"journal":{"ja":"Proc. 12th International Workshop on Frontiers in Handwriting Recognition (ICFHR2010)"},"num_pages":"6","year":2010,"month":11,"location":"Kolkata, India","review":"reviewed","fields":["document-image-analysis"],"doi":"10.1109/ICFHR.2010.127"},{"key":"Sperber2010_ICFHR","id":"1176","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Matthias Sperber","en":""},{"ja":"Martin Klinkigt","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Benjamin Adrian","en":""},{"ja":"Andreas Dengel","en":""}],"title":{"ja":"Handwriting Reconstruction For a Camera Pen Using Random Dot Patterns"},"journal":{"ja":"Proc. 12th International Workshop on Frontiers in Handwriting Recognition (ICFHR2010)"},"num_pages":"6","year":2010,"month":11,"location":"Kolkata, India","review":"reviewed","fields":["document-image-analysis"],"doi":"10.1109/ICFHR.2010.32"},{"key":"Mutoh2010_TENCON","id":"1186","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"","en":"Tomoyuki Mutoh"},{"ja":"","en":"Masakazu Iwamura"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"Theoretical Analysis on Pruning Nearest Neighbor Candidates by Locality Sensitive Hashing"},"journal":{"en":"Proc. 2010 IEEE Region 10 Conference (TENCON2010)"},"num_pages":"4","year":2010,"month":11,"location":"Fukuoka, Japan","review":"reviewed","fields":["nearest-neighbor-search"],"doi":"10.1109/TENCON.2010.5686450","abstract":{"en":"Locality Sensitive Hashing (LSH) is one of the most popular methods of the approximate near neighbor search. In applications that require the nearest neighbors of queries in a short time, LSH is sometimes used in pruning of the candidates of nearest neighbors. While the pruning reduces the processing time greatly, it also reduces the chances of retrieving the exact nearest neighbors.\nHowever, the pruning of nearest neighbor candidates using LSH has not\nbeen considered theoretically. Thus in this paper, we investigate the pruning effect by deriving the formulae of retrieval accuracy and computational cost of distance calculation for uniformly distributed data. Furthermore, we make evaluations on the formulae by comparison between simulation results and the theoretical values."}},{"key":"Inoue2010_CJKPR","id":"1168","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Katsufumi Inoue","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Memory Reduction Method for Specific Object Recognition Based on Lossy Encoding of Feature Vectors with a Bloomier Filter"},"journal":{"ja":"Proceedings of the 2nd China-Japan-Korea Joint Workshop on Pattern Recognition (CJKPR2010)"},"pages":"166-171","year":2010,"month":11,"review":"reviewed","fields":["object-recognition","pattern-recognition"],"abstract":{"ja":"Specific object recognition method with the nearest neighbor search of local features needs an immense\nof memory storage to store them for distance calculation since the number of local features is very large. A way to\nsolve this problem is to skip the distance calculation. In this paper, we propose a memory reduction method with a\nBloomier filter which is far memory efficient than hash tables. From the experiments with 55 3D objects and 5,000\nplanar objects, the proposed method was successful to reduce the required memory for specific object recognition\ncompared with the method with a hash table."}},{"key":"Klinkigt2010_CJKPR","id":"1183","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Martin Klinkigt","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Local Configuration of SIFT-like Features by a Shape Context"},"journal":{"ja":"Proceedings of the 2nd China-Japan-Korea Joint Workshop on Pattern Recognition (CJKPR2010)"},"pages":"11-15","year":2010,"month":11,"review":"reviewed","fields":["object-recognition","pattern-recognition"],"abstract":{"ja":"The representation of information plays the key role in searching a document. While for text documents it is simple to extract keywords to find a document, for images this task is more challenging. In this paper we discuss different representations of images, point out their advantages and disadvantages and present our proposed method for stable and reliable information extraction to address the difficulties in object recognition."}},{"key":"Kobayashi2010","id":"1169","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Takuya Kobayashi","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Recognition of Affine Distorted Characters by Using Affine-InvariantLocal Descriptors"},"journal":{"ja":"Proceedings of the 2nd China-Japan-Korea Joint Workshop on Pattern Recognition (CJKPR2010)"},"pages":"74-77","year":2010,"month":11,"location":"Fukuoka, Japan","review":"reviewed","presen_type":"oral","fields":["pattern-recognition"],"abstract":{"ja":"Camera-based character recognition has a possibility to realize a variety of applications and has been\npaid attention. However, character images are often taken from different angles and characters tend to be distorted.\nIn such a case, recognition accuracy becomes worse. Therefore, we must find some methods to deal with this\nproblem. In this paper, we experimentally evaluate some affine-invariant local descriptors by testing how well\nthey can recognize affine distorted characters. As a result, affine-invariant local descriptors are robust to affine\ndistorted characters to some extent."}},{"key":"Sun2010_CJKPR","id":"1182","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Weihan Sun","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Copyright Protection of Line Drawings:Copyrighted Part Detection Using Cascade Classifiers"},"journal":{"ja":"Proceedings of the 2nd China-Japan-Korea Joint Workshop on Pattern Recognition (CJKPR2010)"},"pages":"9-10","year":2010,"month":11,"review":"reviewed","fields":["pattern-recognition"],"abstract":{"ja":"To protect the copyright of line drawings, we proposed an image retrieval method to detect both printed\nand handwritten partial copies. However, the enormous volumes of line drawing publications with copyright\ncause the problem of scalability. Since not the whole image requires copyright protection, a practical method\nshould detect the copyrighted parts only. In this paper, we propose applying object detection technique to detect\ncopyrighted parts. By using cascade classifiers, we achieved the detection of comic characters’ faces, which are\nimportant copyrighted parts of line drawing publications. The experimental results show the effectiveness of this\nmethod."}},{"key":"Mutoh2010_CJKPR","id":"1173","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Tomoyuki Mutoh","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Derivation of Theoretical Formulas of Accuracy on Accessing Neighboring Buckets in Hash-Based Approximate Nearest Neighbor Search"},"journal":{"ja":"Proceedings of The 2nd China-Japan-Korea Joint Workshop on Pattern Recognition (CJKPR2010)"},"pages":"31-36","year":2010,"month":11,"location":"Fukuoka, Japan","review":"not_reviewed","fields":["nearest-neighbor-search"],"abstract":{"ja":"Approximate nearest neighbor search is a technique which greatly reduces processing time and required amount of memory. Generally, there are the relationships of trade-off among accuracy, processing time and memory amount. Therefore, analysis on the relationships is an important task for practical application of the approximate nearest neighbor search method. In this paper, we construct a model of approximate nearest neighbor search methods with accessing neighboring buckets, and derive theoretical formulas in accuracy. The effectiveness of the formulas have been proved by comparing simulation results with experimental results."}},{"key":"Sakata2010","id":"1171","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Tomohiro Sakata","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Flower Classification by Using Multiple Kernel Learning"},"journal":{"ja":"Proceedings of The 2nd China-Japan-Korea Joint Workshop on Pattern Recognition (CJKPR2010)"},"pages":"144-147","year":2010,"month":11,"review":"reviewed","fields":["pattern-recognition","object-recognition"],"abstract":{"ja":"Object classification for categories with a significant visual similarity is a difficult problem.\nBecause natural objects are slightly different for each individual, it is difficult to classify them with one feature.\nTherefore multiple features are needed for classification.\nAs a method of combining multiple features, MKL has recently been focused.\nIn this research, we employ color, shape, and texture features.\nWe classify the flower images by using MKL and investigate the recognition rate.\nAs a result, the best recognition rate is 75.66% in combining three features with flower 17 category dataset published by Visual Geometry Group of Oxford University."}},{"key":"Takeda2010","id":"1170","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Kazutaka Takeda","en":""},{"ja":"koichi Kise","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Multilingual Document Image Retrieval Based on a Large-Scale Database"},"journal":{"ja":"Proceedings of The 2nd China-Japan-Korea Joint Workshop on Pattern Recognition (CJKPR2010)"},"pages":"42-46","year":2010,"month":11,"location":"Fukuoka, Japan","review":"reviewed","presen_type":"poster","fields":["document-image-analysis","document-image-retrieval"],"awards":{"en":["CJKPR Best Poster Award Winner"]},"abstract":{"ja":"The document image retrieval method that can support various languages has already been proposed.\nHowever, the application of this method for a large-scale database consisting of documents in various languages\nhas not been shown yet. In this paper, we experimentally evaluate the effectiveness of this method for a large-scale\ndatabase of documents in Japanese, Chinese and Korean. From the experimental results, we have confirmed that\nthis method realizes more than 93% accuracy on a 10,000 pages database for each language."}},{"key":"Yoshida2010_CJKPR","id":"1172","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Akira Yoshida","en":""},{"ja":"Marcus Liwicki","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Augmented Handwritings by Data-Embedding Pen"},"journal":{"ja":"Proceedings of The 2nd China-Japan-Korea Joint Workshop on Pattern Recognition (CJKPR2010)"},"pages":"213-215","year":2010,"month":11,"location":"Fukuoka, Japan","review":"reviewed","fields":["document-image-analysis"],"abstract":{"ja":"In this paper, we propose a data-embedding pen, which adds an ink-dot sequence along a handwritten\npattern during writing. The ink-dot sequence represents some meta-information, such as writer’s name, date\nof writing, and URL. The meta-information can be extracted from the handwritten pattern by image processing\ntechniques and a stroke recovery technique. Consequently, we can augment the handwritten pattern by the\ndata-embedding pen to carry arbitrary information."}},{"key":"Kojima2010","id":"1212","type":"intl_conf","language":"EN","show":"show_except_abstract","authors":[{"ja":"Atsuhiro Kojima","en":""},{"ja":"Hiroshi Miki","en":""},{"ja":"Kochi Kise","en":""}],"title":{"ja":"Object Recognition Based on n-gram Expression of Human Actions"},"journal":{"ja":"Proc. of 20th Int. Conf. on Pattern Recognition (ICPR2010)"},"pages":"372-375","year":2010,"month":8,"review":"reviewed","fields":[]},{"key":"Utsumi2010_HBU","id":"1283","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"","en":"Yuzuko Utsumi"},{"ja":"","en":"Yoshio Iwai"},{"ja":"","en":"Hiroshi Ishiguro"}],"title":{"en":"Face Tracking and Recognition Considering the Camera’s Field of View"},"journal":{"en":"Proceedings of International Workshop on Human Behaviour Understanding 2010"},"pages":"52-63","year":2010,"month":8,"location":"Istanbul, Turkey","review":"reviewed","fields":[]},{"key":"Iwamura2010_DAS","id":"1136","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Tomohiko Tsuji","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Memory-Based Recognition of Camera-Captured Characters"},"journal":{"ja":"Proceedings of the 9th IAPR International Workshop on Document Analysis Systems (DAS2010)"},"pages":"89-96","year":2010,"month":6,"location":"Boston, Massachusetts, USA","review":"reviewed","fields":["document-image-analysis","pattern-recognition"],"doi":"10.1145/1815330.1815342","awards":{"en":["IAPR Nakano Award (Best Paper Award) Winner"]},"abstract":{"ja":"This paper addresses how to quickly recognize a characterpattern using a lot of case examples without learning. Herewithout learning means just finding the most similar examplefrom the case examples, and pretend as if the OCR understandsthe definition of the character. This strategy isexpected to work well in most cases with a large dataset,however, also expected to take a lot of time for finding themost similar example. In this paper, we show that a lotof case examples can be processed in a short time. As atestbed, we handle recognition problem of camera-capturedprinted characters. Using a database storing 100 fonts, theproposed method achieved 97.0% of recognition rate for imagescaptured from the right angle and 95.8% for those from45 deg. with 4.56ms of processing time, that is about 220characters per second including everything."}},{"key":"Kise2010_DAS","id":"1154","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Megumi Chikano","en":""},{"ja":"Kazumasa Iwata","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Shinichiro Omachi","en":""}],"title":{"ja":"Expansion of Queries and Databases for Improving the Retrieval Accuracy of Document Portions — An Application to a Camera-Pen System"},"journal":{"ja":"Proceedings of the 9th IAPR International Workshop on Document Analysis Systems (DAS2010)"},"pages":"309-316","year":2010,"month":6,"location":"Boston, Massachusetts, USA","review":"reviewed","fields":["document-image-analysis","document-image-retrieval"],"doi":"10.1145/1815330.1815370","abstract":{"ja":"This paper presents a method of improving the accuracy of document image retrieval focusing on the application to a camera-pen system. In a camera-pen system, document image retrieval is employed for locating the pen-tip position on a page. A serious problem is that since the camera is mounted close to the pen-tip, the camera captures only a tiny portion of the page and the resultant image is under severe perspective distortion, resulting in lowering the retrieval accuracy. To solve this problem, we propose new geometrically invariant features as well as expansion techniques which increase the number of index features of either the database or the query images. From the experimental results, it has been found that the query expansion technique with features by combining affine and perspective invariants allows us the best performance that improves the accuracy of a baseline method more than 27%."}},{"key":"Liwicki2010_DAS","id":"1160","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Marcus Liwicki","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Data-Embedding Pen — Augmenting Ink Strokes with Meta-Information"},"journal":{"ja":"Proceedings of the 9th IAPR International Workshop on Document Analysis Systems (DAS2010)"},"pages":"43-52","year":2010,"month":6,"location":"Boston, Massachusetts, USA","review":"reviewed","fields":["document-image-analysis"],"doi":"10.1145/1815330.1815336"},{"key":"Inoue2010_Conf","id":"1178","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"井上 勝文","en":"Katsufumi Inoue"},{"ja":"黄瀬 浩一","en":"Kochi Kise"}],"title":{"ja":"大規模特定物体認識におけるBloomier Filterを用いたメモリ削減法の有効性の検証"},"journal":{"ja":"平成22年 電気関係学会関西連合大会 講演論文集"},"presen_id":"4P303-15","pages":"922-925","year":2010,"month":11,"review":"not_reviewed","fields":["object-recognition"],"awards":{"ja":["電気関係学会関西連合大会奨励賞受賞"]}},{"key":"Klinkigt2010_Conf","id":"1184","type":"domestic_conf","language":"EN","show":"show","authors":[{"ja":"Martin Klinkigt","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Object Recognition Under Difficult Conditions Based on Superpixel"},"journal":{"ja":"平成22年 電気関係学会関西連合大会 講演論文集"},"presen_id":"4P303-14","pages":"920-921","year":2010,"month":11,"review":"not_reviewed","fields":["object-recognition"],"abstract":{"ja":"In computer vision the task of object recognition is to recognize a certain object in an provided image. For this task a description about the object of interest is learned from images. This process often involves the use of local features like SIFT which can be extracted reliably from images, even if the resolution or the lighting conditions change. The drawback of such local features are that they may lose discriminative power to distinguish between similar objects or the object from the background."}},{"key":"Sun2010_Conf","id":"1180","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"孫 維瀚","en":"Weihan Sun"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"線画の著作権を保護するための類似した部分的複製検出"},"journal":{"ja":"平成22年電気関係学会関西支部連合大会 講演論文集"},"pages":"897-898","year":2010,"month":11,"review":"not_reviewed","fields":["document-image-analysis","pattern-recognition"],"abstract":{"ja":"本稿では，物体認識の手法を用いて，保護する必要がある部分(注目領域) を抽出し，特徴の照合によって，類似した部分の検索法を提案する．実験では，線画出版物の中の最も重要な一つの漫画をデータとして，漫画人物の顔を注目領域として，類似した漫画人物の顔検出実験を行った．その結果より，提案手法の有効性を検証した．"}},{"key":"Sasano2010","id":"1207","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"笹野 泰正","en":"Yasumasa SASANO"},{"ja":"小島 篤博","en":"Atsuhiro KOJIMA"},{"ja":"黄瀬 浩一","en":"Koichi KISE"}],"title":{"ja":"姿勢特徴のn-gram表現を用いた人物行動による物体認識","en":"Object Recognition Based on Human Activity Using Pose n-grams"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"110","number":"296","presen_id":"PRMU2010-114","pages":"7--12","year":2010,"month":11,"review":"not_reviewed","fields":["object-recognition","pattern-recognition"]},{"key":"Yoshida2010","id":"1167","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"吉田 晃","en":""},{"ja":"Marcus Liwicki","en":""},{"ja":"内田 誠一","en":""},{"ja":"岩村 雅一","en":""},{"ja":"大町 真一郎","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"情報埋め込みペンの試作"},"book_title":"平成22年度電気関係学会九州支部連合大会","presen_id":"09-2A-04","pages":"413","year":2010,"month":9,"review":"not_reviewed","fields":[]},{"key":"Barrat2010","id":"1204","type":"domestic_conf","language":"EN","show":"show","authors":[{"ja":"Sabine Barrat","en":""},{"ja":"Geoffrey Alary","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Shinichiro Omachi","en":""}],"title":{"ja":"Is quantity better than quality? — A study case on a simple classifier applied to digit character recognition —"},"journal":{"ja":"Technical Report of IEICE"},"presen_id":"PRMU2010-65","year":2010,"month":9,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"This paper deals with a learning method for digit character recognition.\nWe propose to expand character image databases, by automatically\ngenerating deformations, noises and new fonts, and then to use a simple\nclassifier based on k nearest neighbor technique.\nThe results of first experiments show that the classifier is more robust\non the resulting larger database.\nMoreover, performance on the MNIST digit character database is very\npromising."}},{"key":"Klinkigt2010_MIRU","id":"1185","type":"domestic_conf","language":"EN","show":"show","authors":[{"ja":"Martin Klinkigt","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Using a Reference Point for Local Configuration of SIFT-like Features"},"journal":{"ja":"Proc. of Meeting on Image Recognition and Understanding (MIRU)"},"presen_id":"IS2-29","pages":"960-967","year":2010,"month":7,"review":"not_reviewed","fields":["object-recognition"],"abstract":{"ja":"In this paper we propose a model to describe the local configuration of SIFT-like features. This information helps to overcome the problem with background clutter in images, i.e., information extracted from areas in the image, which do not belong to the object. Our model is not static as many other proposed models. Instead we calculate during detection a reference point of the object and verify this with the features extracted from the query image. This additional step increases the detection performance around 6\\% even under hard conditions, for example, if training images and query images contain background clutter."}},{"key":"Utsumi2010_MIRU","id":"1293","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"内海ゆづ子","en":""},{"ja":"岩井儀雄","en":""},{"ja":"石黒浩","en":""}],"title":{"ja":"確率モデルに基づく動画像の顔追跡・認識"},"journal":{"ja":"画像の認識・理解シンポジウム（MIRU2010）"},"pages":"929-936 (IS2-25)","year":2010,"month":7,"review":"not_reviewed","fields":[]},{"key":"Chikano2010_MIRU","id":"1166","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"近野 恵","en":"Megumi CHIKANO"},{"ja":"岩田 和将","en":"Kazumasa IWATA"},{"ja":"黄瀬 浩一","en":"Koichi KISE"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"},{"ja":"内田 誠一","en":"Seiichi UCHIDA"},{"ja":"大町 真一郎","en":"Shinichiro OMACHI"}],"title":{"ja":"カメラペンシステムにおける射影歪みを考慮した文書画像検索の精度向上法","en":"Accuracy Improvement of Document Image Retrieval on a Camera-Pen System by Coping with Projective Distortion"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2010)論文集","en":"Proceedings of MIRU 2010"},"presen_id":"IS1-30","pages":"239-246","year":2010,"month":7,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"]},{"key":"Iwamura2010_MIRU","id":"1177","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"辻 智彦","en":"Tomohiko Tsuji"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"カメラで撮影した単語画像の実時間認識","en":"Real-Time Recognition of Camera-Captured Word Images"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2010)論文集","en":"Proceedings of MIRU 2010"},"presen_id":"IS1-31","pages":"247-254","year":2010,"month":7,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"本稿ではカメラで撮影した単語画像の実時間認識手法を提案する．\n提案手法を用いることで，\n環境中の単語を認識し，あらかじめ単語に関連付けられた\n情報（画像，音声，翻訳など）を利用者に提供する\n情報提示システムが実現できる．\nこのシステムはノートパソコンで実時間で動作し，\n斜めから撮影した文字列も認識可能である．\nまた，文字が曲線状に配置されていても，\n単語内で文字の大きさが変化しても認識可能である．","en":"In this report, we propose a real-time recognition method of\ncamera-captured word images.\nUsing the method, we can realize a system which recognizes words in\nscenes and offers related information to the word such as images, sounds and\ntranslation of the words in real-time.\nThe system can recognize character images captured from various angles,\ntexts aligned on a non-straight line, and texts which consists of\ncharacters of various sizes."}},{"key":"Kashiwagi2010_IEICEPRMU","id":"1157","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"柏木 隆宏","en":"Takahiro Kashiwagi"},{"ja":"外山 託海","en":"Takumi Toyama"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"局所特徴量の部分空間を用いた3 次元物体認識におけるメモリ削減手法","en":"Memory Reduction Method Using Subspaces of Local Features for 3D Object Recognition"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"Technical Report of IEICE"},"volume":"110","number":"26","presen_id":"PRMU2010-19","pages":"97-102","year":2010,"month":5,"review":"not_reviewed","fields":["object-recognition"],"abstract":{"ja":"3次元物体の認識手法として，画像の局所特徴量を用いた手法が既に提案されている．\nこの手法には，多数の局所特徴量をデータベースに保存しておくため，莫大なメモリ容量が必要となるという問題点がある．\nそこで本稿では，高い認識率を維持しながら，メモリ容量を削減するために，複数の局所特徴量をその部分空間を用いて一つにまとめる手法を提案する．\n提案手法で実験を行った結果，全ての局所特徴量を用いる場合と比べてメモリ容量を約1/18に削減することができ，98.9%の認識率となった．\nさらにこの手法では，認識時に物体の角度の情報を考慮することで，撮影された物体の向きをも検出することができた．"}},{"key":"Toyama2010","id":"1162","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"外山 託海","en":"Takumi Toyama"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"一般物体認識における学習データセットのサイズ拡張による影響の解析","en":"An Analysis of the Impact of a Training Dataset Expansion for GenericObject Recognition"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"110","number":"26","presen_id":"PRMU2010-28","pages":"145-150","year":2010,"month":5,"review":"not_reviewed","fields":["pattern-recognition","object-recognition"],"abstract":{"ja":"一般にパターン認識において，学習データセットの内容は識別性能に大きく影響することが知られている．\nこれまで一般物体認識の分野において用いられてきた多くのデータセットでは，１カテゴリあたりのサンプル数が幾\n分か少ない場合が多いため，データセットを拡張して内容を充実させることができれば認識性能をさらに上げること\nができると考えられる．これまでそのようなサンプル数を増やすための手法はいくつか提案されてはいるが，それら\nの手法で用いられるデータの数は数万程度であり，さらにデータを増やすことで認識性能はより向上すると考えられ\nる．そこで我々はインターネットから数十万枚の画像を集め，それらの画像をフィルタリングしてデータセットに加\nえ，データセットのサイズを大きくした場合とそのままの大きさの場合とで認識性能を比べる．その際識別器を変え\nた上でも実験を行い，それぞれどのような影響を及ぼすかを解析する．","en":"In the field of pattern recognition, it is well known that the contents of a training dataset affects the\nrecognition performance. Most datasets used in the field of generic object recognition have a somewhat small number\nof samples per category. Thus, it is reasonable to think that we can acquire higher recognition performance by\nexpanding the dataset. Although several approaches have attempted to increase the number of samples, the amount\nof these increased samples is about ten thousands which can be increased more to acquire much higher recognition\nperformance. Thus, we collect hundred thousands of images and compare the recognition performance between the\ndataset with original size and the dataset which is expanded with our dataset expansion method. In addition, we\nalso analyze the impact of using different classifiers for these expanded datasets."}},{"key":"Chikano2010_IEICEPRMU","id":"1137","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"近野 恵","en":"Megumi Chikano"},{"ja":"岩田 和将","en":"Kazumasa Iwata"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"}],"title":{"ja":"データベースおよびクエリの拡張による文書画像検索の精度向上法","en":"Improvement of Accuracy of Document Image Retrieval by Expanding Queryies and Databases"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"109","number":"470","presen_id":"PRMU2009-319","pages":"507-512","year":2010,"month":3,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"],"abstract":{"ja":"本稿では，我々が開発中のカメラペンシステムで使用する，文書画像検索の精度向上法を提案する．カメラペンシステムとは，カメラを取り付けたペンで書いた印刷文書への筆跡を，デジタルデータとして復元するシステムである．文書画像検索はデータベースから撮影画像の対応範囲を検索することで，筆跡を求める役割を担っており，高精度での検索が必要不可欠である．しかし現状では，撮影画像に生じる射影歪みの影響で検索精度が低下している．そこで，この問題を解決するために，検索に用いる特徴量の改良，および射影歪みを発生させた画像を生成し，データベースを拡張する手法と，撮影画像を基にクエリを拡張して検索する手法の3つを提案する．有効性を検証するため，改良した特徴量と拡張手法をそれぞれ組み合わせ，改良前の従来手法と比較実験を行った．その結果，改良した特徴量を用い，クエリを拡張した手法を用いたときに，最も精度が向上した．","en":"In this report, we propose a method to improve accuracy of document image retrieval for a camera-pen system. The system is to acquire handwriting on a printed document as digital ink. In this system, document image retrieval is employed to locaate the pen-tip position. Features calculaated based on the foreground image are used to retrieve the document image and the pen-tip position on it. A problem of this system is that severe perspective distortion in the query image prevents us from acquiring an accurate position. To solve this problem, we improve the discrimination power of the features using a perspective invariant. In addition, we propose two expansion methods which bring database image and query image closer. Database expansion is to store geometrically distorted images in the database, and query expansion is to generate transformed images from a query image. From the experimental results, we confirm that the best combination is the query expansion with the proposed feature."}},{"key":"Sperber2010_IEICEPRMU","id":"1149","type":"domestic_conf","language":"EN","show":"show","authors":[{"ja":"Matthias Sperber","en":""},{"ja":"Martin Klinkigt","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Benjamin Adrian","en":""},{"ja":"Andreas Dengel","en":""}],"title":{"ja":"Position Detection For a Camera Pen Using LLAH and Dot Patterns"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"presen_id":"PRMU2009-318","pages":"501-506","year":2010,"month":3,"review":"not_reviewed","fields":["document-image-analysis"],"abstract":{"ja":"In this report, we present a method of position detection for a low-cost camera-pen. Our method allows reconstruction of handwriting, as well as retrieval of the particular document that is being written onto. For this matter, we use a point pattern printed onto the document's background, which is indexed using Locally Likely Arangement Hashing. For position detection, the dot pattern is extracted from the camera image and then matchied to the hash table. Experimental results shows high retrieval accuracy (80.4%--100%) when the pattern is disturbed by the document's foreground only to a certain extent."}},{"key":"Liwicki2010_IEICEPRMU","id":"1159","type":"domestic_conf","language":"EN","show":"show_except_datafile","authors":[{"ja":"","en":"Marcus Liwicki"},{"ja":"","en":"Seiichi Uchida"},{"ja":"","en":"Masakazu Iwamura"},{"ja":"","en":"Shinichiro Omachi"},{"ja":"","en":"Koichi Kise"}],"title":{"en":"An Interface for Embedding Online Information During Writing"},"journal":{"en":"IEICE Technical Report"},"volume":"109","number":"418","presen_id":"PRMU2009-225","pages":"105-110","year":2010,"month":2,"review":"not_reviewed","fields":[]},{"key":"Tada2010_FDEI","id":"1139","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"多田 匡志","en":"Masashi Tada"},{"ja":"武藤 大志","en":"Tomoyuki Muto"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"近さの多段階表現に基づく近似最近傍探索の一般的な分布への拡張","en":"Extensions of Approximate Nearest Neighbor Search Based on a Multi-Valued Expression of Closeness to General Distributions"},"journal":{"ja":"データ工学と情報マネジメントに関するフォーラム論文集","en":"Proc. First Forum on Data Engineering and Information Management"},"presen_id":"F3-2","year":2010,"month":2,"fields":["nearest-neighbor-search"],"abstract":{"ja":"近似最近傍探索は，\nクエリと最も距離が近い点を探索する最近傍探索問題において，\n探索精度を犠牲にすることで計算時間，メモリ使用量を大幅に削減する手法である．\n我々は既に2つの近似最近傍探索手法MVH1とMVH2を提案した．\nこれらの手法は距離尺度がL1ノルムであり，データの分布が一様分布の\n場合で有効性を確認した．\n本稿では，これらの手法の制約を排除し，Lpノルムならびに\n一般的な分布に適用可能な手法を提案する．\n距離尺度がL2ノルムである時，改良した１つ目の提案手法は，計算時間をLSHの約70%に削減した．\n改良した２つ目の提案手法は，LSHに対してメモリを約50%削減した．","en":"Approximate nearest neighbor search is a technique which greatly\nreduces processing time and required amount of memory\nfor nearest neighbor search.\nWe proposed two approximate nearest neighbor methods: MVH1 and MVH2.\nWe comfirmed the effectiveness of these methods for uniform distribution with L1 norm.\nIn this report, we get rid of the restrictions of these methods and propose efficient methods for Lp norm and general distributions.\nFor L2 norm, improved MVH1 achived about 75% of processing time of LSH.\nImproved MVH2 achived about 50% of required amount of memory of LSH."}},{"key":"Iwamura2010_IEICEPRMU","id":"1135","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu IWAMURA"},{"ja":"辻 智彦","en":"Tomohiko TSUJI"},{"ja":"黄瀬 浩一","en":"Koichi KISE"}],"title":{"ja":"カメラ撮影文字の事例に基づく実時間認識","en":"Real-Time Case-Based Recognition of Camera-Captured Characters"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"109","number":"418","presen_id":"PRMU2009-222","pages":"87-92","year":2010,"month":2,"review":"not_reviewed","fields":["document-image-analysis","pattern-recognition"],"note":{"ja":"PRMUの原稿に誤りがあったため、訂正版を置いてあります。下記が大きな修正点です．\n\n<ul>\n<li>図12(a)<br />\n従来手法の認識率が間違っていて，本来の値より12%くらい低くなっていた．そのため，提案手法で上昇した認識率は約20%から約8%になった．</li>\n\n<li>3節の2段落目<br />\n提案した3つの方策のうち，1番目のものの閾値の設定方法が間違っていた．</li>\n\n<li>図9<br />\n図中の「生成された二値ベクトル」が5本になっていたが，正しくは3本だった．</li>\n</ul>"},"abstract":{"ja":"本稿では，カメラで撮像した文字の認識問題において，事例を大量にデータベースに蓄積しておくことで，高速かつ正確な認識を実現する方法を述べる．具体的には，我々が既に提案している実時間カメラベース文字認識手法に，我々が特定物体認識のために提案している高速かつ頑健な近似最近傍探索手法を新たに導入する．これにより，100 種類のフォントを学習しても認識率と処理時間の両方において高い性能を実現することが確認できた．","en":"In this paper, in the recognition problem of camera-captured characters, we present how to realize quickand precise recognition using a lot of case examples stored in a database. To be more precise, we introduce our fastand robust approximate nearest neighbor search method, originally designed for speci\fc object recognition, to ourreal-time camera-based character recognition method. In the experiments using a database storing 100 fonts, wecon\frmed that the proposed method achieved high performance in both recognition rate and processing time."}},{"key":"Iwata2010_IEICEPRMU","id":"1138","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩田 和将","en":"Kazumasa Iwata"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"}],"title":{"ja":"特徴追跡と文書画像検索を用いたカメラペンシステム","en":"Camera Pen System Using Feature Tracking and Document Image Retrieval"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"109","number":"418","presen_id":"PRMU2009-214","pages":"39-44","year":2010,"month":2,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"],"abstract":{"ja":"本研究では，ペンにカメラを搭載し画像処理をすることで，筆跡をデジタルデータとして取得する手法を提案する．\n筆記には，単純な白紙へのメモ書きや，文書への下線などの文書に情報を加えるものがある．\nそこで，白紙において筆跡の復元可能で，かつ文書領域では筆記対象の文書名や文書中の位置を特定できる必要がある．\nこの要求を実現するため，本研究では，既に提案されている白紙に対応する手法と文書画像検索が可能な手法を統合する．\n統合の際の大きな問題は，両手法で処理に必要な撮影範囲が大幅に異なる点（前者は狭く，後者は広い）である．\nこの問題に対処するため，提案手法では画像モザイキングを導入する．筆跡の復元は，前者の手法により紙の凹凸や印刷文字から得られるSURF特徴点の追跡により実現する．\n同時に，SURF特徴点を用いたモザイキングにより画像を貼り合わせる．\n十分な大きさの領域が得られると，後者の手法により文書画像を検索し筆跡を文書に対応付ける．本稿では，復元された筆記を実験により評価するとともに，今後の課題についても考察する．"}},{"key":"Marcus2010","id":"1799","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"マーカス リビキ","en":"LIWICKI Marcus"},{"ja":"内田 誠一","en":"UCHIDA Seiichi"},{"ja":"岩村 雅一","en":"IWAMURA Masakazu"},{"ja":"大町 真一郎","en":"OMACHI Shinichiro"},{"ja":"黄瀬 浩一","en":"KISE Koichi"}],"title":{"ja":"データ埋め込みペンの実装","en":"An Interface for Embedding Online Information During Writing"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE technical report"},"volume":"109","number":"418","presen_id":"PRMU2009-225","pages":"105-110","year":2010,"month":2,"review":"not_reviewed","fields":["document-image-analysis"]},{"key":"Tajima2010","id":"1158","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"田島修司","en":""},{"ja":"内田誠一","en":""},{"ja":"岩村雅一","en":""},{"ja":"大町真一郎","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"ペン先カメラ画像と紙面上特徴点を用いた手書きパターンの復元"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"109","number":"418","presen_id":"PRMU2009-208","pages":"7-12","year":2010,"month":2,"review":"not_reviewed","fields":[]},{"key":"Tsuji2010_IEICEPRMU","id":"1134","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"辻 智彦","en":"Tomohiko TSUJI"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"},{"ja":"黄瀬 浩一","en":"Koichi KISE"}],"title":{"ja":"リアルタイム単語認識技術を利用したカメラベース情報取得システム","en":"A Camera-Based Information Acquisition System Using a Technique of Real-Time Word Recognition"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"109","number":"418","presen_id":"PRMU2009-216","pages":"51-56","year":2010,"month":2,"review":"not_reviewed","fields":["document-image-analysis","pattern-recognition"],"abstract":{"ja":"ユーザーインタフェースの利便性に関わる要素としては素早い応答や直観的な操作性が挙げられる．我々が既に提案しているカメラベース文字認識手法は実時間処理が可能，射影歪みに頑健，文書レイアウトに依存しない，という特徴を持つ．我々はこの手法を利用することで新たな対話型インタフェースが実現するのではないかと考えた．本稿ではカメラで撮影した環境中の文字を認識して関連した情報を提供するシステムを提案し，ノートパソコン上で実時間で動作するデモシステムを紹介する．システムには1 文字単位の認識結果から単語を推定する処理が必要なため，有効な単語認識手法についても述べる．提案手法ではサーバーを用いた場合1 秒間に約42 単語の認識が可能である．","en":"In order to realize a comfortable user interface, rapid response and intuitive operation are required. Thecamera-based character recognition method where we have already proposed has some advantages, which are readyfor real-time processing, robust to perspective distortion, and free from layout constraints of documents. We thinkthat there is a possibility to realize a new interactive interface by applying it. In this paper, we propose a systemwhich recognizes characters captured with a camera and offers related information. We demonstrate a system whichworks on a laptop PC in real-time. Since it needs a process that recognizes words from characters recognized separately,we propose an effective word recognition method for the system. Our proposed method achieved to recognizeabout 42 words per second on a server."}},{"key":"Muto2010","id":"1131","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"武藤 大志","en":"Tomoyuki Muto"},{"ja":"多田 匡志","en":"Masashi Tada"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"ハッシュを利用した近似最近傍探索における隣接バケット参照の精度とメモリ使用量の理論式の導出","en":"Derivation of Theoretical Formulae of Accuracy and Memory Amount on Accessing Neighboring Buckets in Hash-Based Approximate Nearest Neighbor Search"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Tecnical Report"},"volume":"109","number":"374","presen_id":"PRMU2009-170","pages":"109-114","year":2010,"month":1,"fields":["nearest-neighbor-search"],"abstract":{"ja":"近似最近傍探索は，クエリと最も距離が近い点を探索する最近傍探索の\n計算時間，メモリ使用量を大幅に削減する手法である．\n一般に精度，計算時間，メモリ使用量はトレードオフの関係にあり，\nその関係を解析することは，様々な場面に近似最近傍探索を適用する上で，\n重要な課題である．\n本稿では，ハッシュを利用した近似最近傍探索において，\n文献~\\cite{nog,PCH,M-P_LSH,mvh}で行われている“隣接バケットを参照する”\n方策のモデル化を行い，\n精度とメモリ使用量に関して理論式を求める．\nそして，実験とシミュレーションにより理論式の妥当性を検証する．","en":"Approximate nearest neighbor search is a technique which greatly\nreduces processing time and required amount of memory \nfor nearest neighbor search.\nGenerally, there are the relationships of trede-off\namong accuracy, processing time and memory amount. \nThus, analysis on the relationships is an important task for\nactual use of approximate nearest neighbor search method.\nIn this paper, we construct a model of approximate nearest neighbor search methods with\naccessing neighboring buckets ~\\cite{nog,PCH,M-P_LSH,mvh},\nand derive theoretical formulae in accuracy and memory amount.\nWe compare simulated values with experimented values."}},{"key":"Kise2010","id":"1205","type":"book","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"Bag-of-Features に基づく物体認識（１） — 特定物体認識 —"},"book_title":"CVIMチュートリアルシリーズコンピュータビジョン最先端ガイド","volume":"3","pages":"63-84","year":2010,"month":12,"publisher":"アドコム・メディア","review":"not_reviewed","fields":["object-recognition"]},{"key":"Hiroyuki2010","id":"1146","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"住岡 浩之","en":"Sumioka Hiroyuki"}],"title":{"ja":"Visual Word 作成における適応的クラスタリング手法の比較","en":"Comparison of Adaptive Clustering Methods in Making Visual-Word"},"year":2010,"month":3,"review":"not_reviewed","fields":["object-recognition"]},{"key":"Inoue2010_MThesis","id":"1143","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"井上 勝文","en":"Katsufumi Inoue"}],"title":{"ja":"特定物体認識におけるBloomier Filterを用いた特徴ベクトルの非可逆符号化によるメモリ削減法","en":"Memory Reduction Method for Specific Object Recognition Based on Lossy Encoding of Feature Vectors with the Bloomier Filter"},"year":2010,"month":3,"review":"not_reviewed","fields":["object-recognition"]},{"key":"Iwata2010_MThesis","id":"1148","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"岩田 和将","en":"Kazumasa Iwata"}],"title":{"ja":"特徴追跡と文書画像検索を用いたカメラペンシステム","en":"Camera Pen System Using Feature Tracking and Document Image Retrieval"},"year":2010,"month":3,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"]},{"key":"Miki2010","id":"1147","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"三木 博史","en":"Hiroshi Miki"}],"title":{"ja":"物体認識のためのn-gramを用いた人物動作表現に関する研究","en":"A Research on Representation of Human Action using n-grams for Object Recognition"},"year":2010,"month":3,"review":"not_reviewed","fields":["object-recognition"]},{"key":"Sun2010_MThesis","id":"1144","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"孫 維瀚","en":"Weihan Sun"}],"title":{"ja":"線画の著作権保護のための部分的複製検出法に関する研究","en":"Study on Partial Copy Detection for Copyright Protection of Line Drawings"},"year":2010,"month":3,"review":"not_reviewed","fields":["document-image-analysis","information-retrieval","pattern-recognition"]},{"key":"Tada2010_MThesis","id":"1140","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"多田 匡志","en":"Masashi Tada"}],"title":{"ja":"近さの多段階表現に基づく近似最近傍探索","en":"Approximate Nearest Neighbor Based on a Multi-Valued Expression of Closeness"},"year":2010,"month":3,"fields":["nearest-neighbor-search"]},{"key":"Tsuji2010_MThesis","id":"1142","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"辻 智彦","en":"Tomohiko Tsuji"}],"title":{"ja":"カメラで撮像した環境中文字列の実時間認識システム","en":"A Real-Time Recognition System ofScene Texts Captured by a Camera"},"year":2010,"month":3,"review":"not_reviewed","fields":["document-image-analysis","pattern-recognition"]},{"key":"Asada2010","id":"1164","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"浅田伸彦","en":""}],"title":{"ja":"カラー文字領域の高速な切り出しの検討"},"year":2010,"month":3,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Chikano2010_BThesis","id":"1150","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"近野 恵","en":""}],"title":{"ja":"カメラペンシステムにおける文書画像検索の精度向上法"},"year":2010,"month":3,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"]},{"key":"Kashiwagi2010_BThesis","id":"1163","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"柏木 隆宏","en":""}],"title":{"ja":"3次元物体認識における局所特徴量の部分空間を用いたメモリ容量の削減"},"year":2010,"month":3,"review":"not_reviewed","fields":["object-recognition"]},{"key":"Ueda2010","id":"1161","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"上田敬介","en":""}],"title":{"ja":"部分盗用検出におけるメモリ削減"},"year":2010,"month":3,"review":"not_reviewed","fields":["information-retrieval"]},{"key":"Kise2010_J_Specific","id":"1187","type":"survey","language":"JA","show":"show_except_datafile","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"局所特徴量を用いた画像照合による特定物体認識","en":"Specific Object Recognition by Image Matching Using Local Features"},"journal":{"ja":"人工知能学会誌","en":"Journal of the Japanese Society of Artificial Intelligence"},"volume":"25","number":"6","pages":"769-776","year":2010,"month":11,"review":"not_reviewed","fields":["object-recognition"]},{"key":"Uchida2010_ITE","id":"1141","type":"survey","language":"JA","show":"show_except_datafile","authors":[{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"Marcus Liwicki","en":"Marcus Liwicki"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"ディジタルペン","en":"Digital Pen"},"journal":{"ja":"映像情報メディア学会誌","en":"ITE Journal"},"volume":"64","number":"3","pages":"293-298","year":2010,"month":3,"review":"not_reviewed","fields":["document-image-analysis"],"abstract":{"ja":"ディジタルペンについて概観する．ディジタルペンは，スムーズな文字・図形入力機能，および迅速なポインティング機能を持った，優れたインタフェースである．種類としては，筆記対象が限定されたもの，筆記対象が任意のものに大別される．本稿では，タブレットやアノトペンなど既に製品化されている技術について述べ，今後の課題を考察する．"}},{"key":"Kise2010_J_Recent","id":"1133","type":"survey","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"講座：第２回　マルチメディア検索の最先端　大規模静止画像DBの検索","en":"Recent Technologies in Multimedia Retrieval (2): Retrieval of Large-Scale Databases of Still Images"},"journal":{"ja":"映像情報メディア学会誌","en":"The Journal of The Institute of Image Information and Television Engineers"},"volume":"64","number":"2","pages":"192-197","year":2010,"month":2,"review":"not_reviewed","fields":["object-recognition"]},{"key":"Iwamura2010_SPIE","id":"1188","type":"other","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Tomohiko Tsuji","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Real-life clickable text"},"journal":{"ja":"SPIE Newsroom"},"year":2010,"month":12,"review":"not_reviewed","fields":["document-image-analysis","pattern-recognition"],"doi":"10.1117/2.1201012.003308","url":"http://spie.org/x43601.xml","abstract":{"ja":"A camera-based interface detects and identifies text in a scene, providing additional relevant information on request."}},{"key":"Noguchi2009_IEICE_Efficient","id":"1101","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"野口 和人","en":"Kazuto NOGUCHI"},{"ja":"黄瀬 浩一","en":"Koichi KISE"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"}],"title":{"ja":"近似最近傍探索の多段階化による高速特定物認識","en":"Efficient Recognition of Specific Objects by Cascading Approximate Nearest Neighbor Searchers"},"journal":{"ja":"電子情報通信学会論文誌D","en":"Trans. IEICE"},"volume":"J92-D","number":"12","pages":"2238-2248","year":2009,"month":12,"review":"reviewed","fields":["object-recognition","nearest-neighbor-search"],"url":"http://search.ieice.org/bin/summary.php?id=j92-d_12_2238","note":{"ja":"論文のPDFファイルを<a href=\"http://www.ieice.org/jpn/trans_online/\">電子情報通信学会 Transactions online</a>からダウンロードすることができます．","en":"PDF file is available at <a href=\"http://www.ieice.org/jpn/trans_online/\">IEICE Transactions online (in Japanese)</a>."},"abstract":{"ja":"SIFTなどの局所特徴量の最近傍探索によって特定物体認識を行う場合には，\n一画像あたりの特徴ベクトルの数が膨大になるため，\n最近傍探索の効率が重要となる．\n本論文では，\n「正しい認識を可能とする最近傍探索の精度は画像によって異なる」\nという観点から処理を削減した効率的認識法を提案する．\n具体的には，近似最近傍探索に基づく識別器を多段階に縦列接続\nすることにより，認識に用いる近似の程度を画像に応じて変更し，\n大幅な効率化を実現する．\n一万枚の平面物体のデータベースを用いた実験の結果，\n処理時間を，多段階化を行わなかった場合の約1/10，\nANNやLSHを近似最近傍探索の手法として用いた場合の約1/40に削減でき，\n例えば，認識率98%，処理時間1ms/queryを達成できることがわかった．\nまた，8.6%のリジェクトを行うことによって，\n誤認識率を0%に抑えることもできた．\nさらに，10万画像を用いた実験から，\n高いスケーラビリティがあることもわかった．","en":"This paper concerns the task of recognizing object instances,\nor specific object recognition, with the special focus on\nplanar object recognition. For this task, it is common to\nemploy local features such as SIFT for indexing images of objects\nand match them using approximate nearest neighbor search for efficiency.\nIn this paper, we propose a new method of more efficient recognition\nby taking into account the fact\nthat the appropriate level of accuracy of nearest neighbor search \ndepends on images to be recognized.\nThe proposed method is characterized by the mechanism\ncalled cascaded recognizers:\nmultiple recognizers with approximate nearest neighbor search\nare cascaded so as to determine automatically the appropriate\nlevel of approximation depending on the input image.\nFrom experimental results with 10,000 images,\nwe have confirmed that the proposed method is capable of \nachieving a recognition rate of 98% in 1 ms / query,\nwhich is 1/10 of the recognition time without the cascade,\nand 1/40 of the recognition time with conventional \napproximate nearest neighbor search such as ANN and LSH.\nIn addition, a recognition error rate of the proposed method \nhas been suppressed to 0% by allowing a rejection rate of 8.6%.\nExperimental results with 100,000 images show high scalability of the \nproposed method."}},{"key":"Noguchi2009_IEICE_Experimental","id":"1087","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"野口 和人","en":"Kazuto NOGUCHI"},{"ja":"黄瀬 浩一","en":"Koichi KISE"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"}],"title":{"ja":"大規模特定物体認識における認識率，処理時間，メモリ量のバランスに関する実験的検討","en":"Experimental Evaluation of Balancing the Recognition Rate, Processing Time and Memory Requirement for Large-Scale Recognition of Specific Objects"},"journal":{"ja":"電子情報通信学会論文誌D","en":"Trans. IEICE"},"volume":"J92-D","number":"8","pages":"1135-1143","year":2009,"month":8,"review":"reviewed","fields":["object-recognition","nearest-neighbor-search"],"url":"http://search.ieice.org/bin/summary.php?id=j92-d_8_1135","note":{"ja":"論文のPDFファイルを<a href=\"http://www.ieice.org/jpn/trans_online/\">電子情報通信学会 Transactions online</a>からダウンロードすることができます．","en":"PDF file is available at <a href=\"http://www.ieice.org/jpn/trans_online/\">IEICE Transactions online (in Japanese)</a>."},"abstract":{"ja":"局所記述子を用いた物体認識において，認識率，処理時間，メモリ量について良いバランスを実現することは重要な問題である．従来，局所記述子をベクトル量子化してvisual word を得，それを認識に用いる手法が多数提案されている．インスタンス(特定物体) を対象とした認識では，数多くのvisual word を用いる手法は高い認識率が得られるものの，処理時間やメモリ量に問題が生じる．一方，局所記述子を量子化せずそのまま用いる手法であっても，近似最近傍探索によって処理が大幅に短縮できることが知られている．本稿では，ベクトル量子化を行わない場合でも，メモリ量の問題を解決する手法として，特徴ベクトルの各次元を記録するビット数を削減する手法を提案する．10 万画像のモデルを用いて認識実験を行った結果，各次元を2 ビットで表現したとき，ビット数削減前(16 ビット) と比べて，認識率は少し下がるものの，処理時間は殆ど変化しないことが分かった．","en":"For object recognition based on local descriptors, balancing the recognition rate, processing time and memory requirement is an important issue. In conventional methods, local descriptors are vector-quantized to “visual words” for describing images. Although many visual words allow us to improve the recognition rate for\nrecognizing instances (specific objects), they also pose a problem of processing time and memory requirement. On the other hand, even if a recognition method employs local descriptors without vector quantization, approximate nearest neighbor search enables us to cut the processing time drastically. In this report, we propose a method of memory reduction applicable even to methods without vector quantization. The proposed method reduces the memory requirement by limiting the number of bits for the record of each dimension of feature vectors. From experimental results on 100,000 images, it has been confirmed that a representation with 2 bit / dimension subtly decreases the recognition rate, but hardly changes the processing time."}},{"key":"Kise2009_DBSJ","id":"1088","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"野口 和人","en":"Kazuto Noguchi"}],"title":{"ja":"局所特徴量のハッシングによる大規模画像検索","en":"Large-Scale Image Retrieval by Hashing of Local Features"},"journal":{"ja":"日本データベース学会論文誌","en":"DBSJ Journal"},"volume":"8","number":"1","pages":"119-124","year":2009,"month":6,"review":"reviewed","fields":["document-image-analysis","document-image-retrieval","object-recognition"],"url":"http://dbsj.org/journal/dbsj_journal/dbsj_journal_vol_8_no_1_119_124/","abstract":{"ja":"検索質問を画像とする画像検索は，バーコードの代替や著作権保護など，数多くのアプリケーションを持つ重要なタスクである．一般にカメラ付き携帯電話などのデバイスで撮影される検索質問は，データベースの画像と同じものを写していても，撮影条件の変動によりデータとしては全く異なるため，検索は容易ではない．この問題は，局所特徴量と呼ばれる撮影条件にあまり影響を受けない特徴量を用いると解決できるが，特徴量の数が膨大となるため，その照合に要する処理時間が新たに問題となる．本論文では，この問題を解決する方策として，我々が取り組んでいるハッシュ表を用いた検索手法について述べる．検索対象としては，風景や人物を撮影した一般の画像に加え，文書画像という全く性質の異なるものを取り上げ，各々に有効な局所特徴量を示すとともに，両者の検索に共通する考え方を述べる．","en":"Image retrieval with query images is a task to find their corresponding images (near duplicates) from the database. It is an important task since it has many applications such as a replacement of barcodes and copyright protection of images. The task of retrieval is not trivial since query images are often far different from their corresponding images in the database due to different imaging conditions. Although this problem can be solved using local features, which are known to be robust to change of the imaging conditions, they pose a problem of processing time since their number is huge. In this paper, we describe our new retrieval methods based on hashing for solving the above problem. As images in the database, we deal with ordinary images of scenes, objects and people, etc., as well as document images that are completely dissimilar to ordinary images. Local features for each type and the common notion for successful retrieval are shown."}},{"key":"Sezaki2009","id":"1076","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"瀬崎 直人","en":"Naoto Sezaki"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"タグの共起と類似画像を用いたタグ付け支援システム","en":"A System for Recommending Tags of Images Using Co-Occurrence of Tags and Similar Images"},"journal":{"ja":"電気学会論文誌(C)","en":"IEEJ Transactions on Electronics, Information and Systems"},"volume":"129","number":"3","pages":"475-480","year":2009,"month":3,"review":"reviewed","fields":["information-retrieval"]},{"key":"Inoue2009_IEICE","id":"1108","type":"letter","language":"JA","show":"show_except_datafile","authors":[{"ja":"井上 勝文","en":"Katsufumi Inoue"},{"ja":"三宅 弘志","en":"Hiroshi Miyake"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"三次元物体認識に用いる局所特徴量の取捨選択によるメモリ使用量削減手法の実験的検討","en":"Experimetal Investigation of a Memory Reduction Method for 3D Object Recognition Based on Selection of Local Features"},"journal":{"ja":"電子情報通信学会論文誌D","en":"Trans. IEICE"},"volume":"J92-D","number":"9","pages":"1686-1689","year":2009,"month":9,"review":"reviewed","fields":["object-recognition","pattern-recognition"],"note":{"ja":"論文のPDFファイルを<a href=\"http://www.ieice.org/jpn/trans_online/\">電子情報通信学会 Transactions online</a>からダウンロードすることができます．"},"abstract":{"ja":"局所特徴量を用いる最も単純な物体認識手法では，大量の局所特徴量を用いモデルを構築し，認識を行うが，局所特徴量を記憶するためにばく大なメモリ容量が必要となる．本論文では，認識に有効な局所特徴量を取捨選択し，メモリ容量を削減する手法を提案する．","en":"PDF file is available at <a href=\"http://www.ieice.org/jpn/trans_online/\">IEICE Transactions online (in Japanese)</a>."}},{"key":"Sun2009_CJKPR","id":"1129","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Weihan Sun","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Speeding up the Detection of Line Drawings Using a Hash Table"},"journal":{"ja":"CJKPR09"},"volume":"2","pages":"896-900","year":2009,"month":11,"review":"reviewed","fields":["document-image-analysis","pattern-recognition"],"abstract":{"ja":"Line drawings, such as comics, marks and logos, are an important part of image publications, which need copyright protection. Since line drawings are composed by lines, it is easy to make a similar copy by handwriting. Therefore, in the case of line drawings, copyright protection method should also be available for handwritten copies. Although a method applicable to handwritten copy detection has been proposed, its processing is time-consuming since it is based on matching a large number of local features extracted from images. In this paper, we apply a hash based method to speed up the detection of line drawings. Compared to the previous method, the detection time per image reduces from 6, 013 ms to 75 ms with almost the same detection rate. Furthermore, we have also proved that the proposed method has effectiveness for detection with a larger database."}},{"key":"Utsumi2009_ICSIPA","id":"1284","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"","en":"Yuzuko Utsumi"},{"ja":"","en":"Yuta Matsumoto"},{"ja":"","en":"Yoshio Iwai"}],"title":{"en":"An Efficient Branch and Bound Method for Face Recognition"},"journal":{"en":"The IEEE International Conference on Signal and Image Processing Applications (ICSIPA)"},"year":2009,"month":11,"location":"Kuala Lumpur, Malaysia","review":"reviewed","fields":[]},{"key":"Utsumi2009_OMNIVIS","id":"1285","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"","en":"Yuzuko Utsumi"},{"ja":"","en":"Yoshio Iwai"},{"ja":"","en":"Hiroshi Ishiguro"}],"title":{"en":"Face Tracking by using Omnidirectional Sensor Network"},"journal":{"en":"Ninth Workshop on Omnidirectional Vision, Camera Networks and Non-classical Cameras (OMNIVIS 2009)"},"pages":"2172-2179","year":2009,"month":10,"location":"Kyoto, Japan","review":"reviewed","fields":[]},{"key":"Inoue2009_WSLAVD","id":"1109","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Katsufumi Inoue","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Compressed Representation of Feature Vectors Using a Bloomier Filter and Its Application to Specific Object Recognition"},"journal":{"ja":"Proceedings of the 1st International Workshop on Emergent Issues in Large Amount of Visual Data (WS-LAVD2009)"},"pages":"2133-2140","year":2009,"month":10,"review":"reviewed","fields":["object-recognition","pattern-recognition"],"abstract":{"ja":"Nearest Neighbor search of feature vectors representing local features is often employed for specific object recognition. In such a method, it is required to store many feature vectors to match them by distance calculation. The number of feature vectors is, in general, so large that a huge amount of memory is needed for their storage. A way to solve this problem is to skip the distance calculation because no feature vectors need to be stored if there is no need to calculate the distance. In this paper, we propose a method of object recognition without distance calculation. The characteristic point of the proposed method is to use a Bloomier filter, which is far memory efficient than hash tables, for storage and matching of feature vectors. From experiments of planar and 3D specific object recognition, the proposed method is evaluated in comparison to a method with a hash table."}},{"key":"Kise2009_WSLAVD","id":"1110","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Kazuto Noguchi","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Robust and Efficient Recognition of Low-quality Images by Cascaded Recognizers with Massive Local Features"},"journal":{"ja":"Proceedings of the 1st International Workshop on Emergent Issues in Large Amount of Visual Data (WS-LAVD2009)"},"pages":"2125-2132","year":2009,"month":10,"location":"Kyoto, Japan","review":"reviewed","fields":["object-recognition"],"doi":"10.1109/ICCVW.2009.5457543","abstract":{"ja":"For image recognition with camera phones,\ndefocus and motion blur cause a serious drop of the image \nrecognition rate.\nIn this paper, we employ generative learning,\ni.e., generating blurred images\nand learning based on massive local features\nextracted from them,\nfor a recognition method using \napproximate nearest neighbor search\nof local features.\nMajor problems of generative learning are\nlong processing time and a large amount of memory required for nearest \nneighbor search.\nThe problems become serious when we use a large-scale database.\nIn the proposed method, they are solved by cascaded recognizers and scalar \nquantization.\nFrom experimental results with up to one million images, we have confirmed that \nthe proposed method improves the recognition rate,\nand cuts the processing time as compared to \na method without generative learning."}},{"key":"Utsumi2009_ICDSC","id":"1286","type":"intl_conf","language":"JA","show":"show","authors":[{"ja":"","en":"Yuzuko Utsumi"},{"ja":"","en":"Yoshio Iwai"}],"title":{"en":"Face Tracking and Recognition by using Omnidirectional Sensor Network"},"journal":{"en":"Proceedings on third ACM/IEEE International Conference on Distributed Smart Cameras (ICDSC 2009)"},"year":2009,"month":9,"location":"Como, Italy","review":"reviewed","fields":[]},{"key":"Sun2009_MITA","id":"1104","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Weihan Sun","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"A Method for Copyright Protection of Line Drawings"},"journal":{"ja":"Proceedings of the 2009 International Conference on Multimedia, Information Technology and its applications (MITA2009)"},"pages":"167-168","year":2009,"month":8,"review":"reviewed","fields":["information-retrieval"]},{"key":"Inoue2009_MITA","id":"1102","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Katsufumi Inoue","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"A Method of Memory Reduction for Specific Object Recognition with a Bloomier Filter"},"journal":{"ja":"Proceedings of the 2009 International Conference on Multimedia, Information Technology and its Applications (MITA2009)"},"pages":"27-28","year":2009,"month":8,"review":"reviewed","fields":["object-recognition","pattern-recognition"],"abstract":{"ja":"Specific object recognition based on nearest neighbor search of feature vectors requires a huge amount of memory to store all feature vectors for distance calculation. To solve this problem, we propose a memory reduction method for specific object recognition with no distance calculation for matching feature vectors. The proposed method is characterized by the use of a Bloomier filter, which is far memory efficient than a hash table, for the storage of feature vectors. The proposed method is evaluated based on experiments of planar and 3D specific object recognition in comparison to a method with a hash table."}},{"key":"Klinkigt2009_MITA","id":"1103","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Martin Klinkigt","en":""},{"ja":"Heiko Maus","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Andreas Dengel","en":""}],"title":{"ja":"Automatic Management of Large Images Databases"},"journal":{"en":"Proceedings of the 2009 International Conference on Multimedia, Information Technology and its Applications (MITA2009)"},"pages":"23-24","year":2009,"month":8,"review":"reviewed","fields":["object-recognition","pattern-recognition"],"abstract":{"ja":"With an increasing amount of images a user would like to keep and share the desire for an adequate management for\nthese images raises. The normal folder structure which is available on nowadays computers can fulfill this task only non satisfying. Often an image contains more individual information than can be expressed by a hierarchical tree.\nWe introduce annotation techniques to support the user in this organization. Furthermore, we analyze how a\ncomputer system could do this organization automatically to keep the amount of work for the user as low as possible."}},{"key":"Toyama2009_MITA","id":"1105","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Takumi Toyama","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Object Recognition Using Hierarchical Structure"},"journal":{"ja":"Proceedings of the 2009 International Conference on Multimedia, Information Technology and its Applications (MITA2009)"},"pages":"41-44","year":2009,"month":8,"review":"reviewed","fields":["object-recognition"]},{"key":"Iwamura2009_ICDAR","id":"1100","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Tomohiko Tsuji","en":""},{"ja":"Akira Horimatsu","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Real-Time Camera-Based Recognition of Characters and Pictograms"},"journal":{"ja":"Proceedings of the 10th International Conference on Document Analysis and Recognition (ICDAR2009)"},"pages":"76-80","year":2009,"month":7,"location":"Barcelona, Spain","review":"reviewed","fields":["document-image-analysis"],"doi":"10.1109/ICDAR.2009.259","abstract":{"ja":"Camera-based character recognition systems should have the capability of quick operation and recognizing perspectively distorted texts in a complex layout. In this paper, in order to realize such a system, we propose a simple but efficient implementation of camera-based recognition of characters and pictograms. With help of new hashing and voting techniques, the proposed method runs well in real-time even on a laptop PC with a web camera."}},{"key":"Iwata2009","id":"1107","type":"intl_conf","language":"JA","show":"show","authors":[{"ja":"Kazumasa Iwata","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Tomohiro Nakai","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Shinichiro Omachi","en":""}],"title":{"ja":"Capturing Digital Ink as Retrieving Fragments of Document Images"},"journal":{"ja":"Proceedings of the 10th International Conference on Document Analysis and Recognition (ICDAR2009)"},"pages":"1236-1240","year":2009,"month":7,"location":"Barcelona, Spain","review":"reviewed","fields":["document-image-analysis","document-image-retrieval"],"doi":"10.1109/ICDAR.2009.192"},{"key":"Nakai2009_ICDAR","id":"1099","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Tomohiro Nakai","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Real-Time Retrieval for Images of Documents in Various Languages using a Web Camera"},"journal":{"ja":"Proceedings of the 10th International Conference on Document Analysis and Recognition (ICDAR2009)"},"pages":"146-150","year":2009,"month":7,"location":"Barcelona, Spain","review":"reviewed","fields":["document-image-analysis","document-image-retrieval"],"doi":"10.1109/ICDAR.2009.125","abstract":{"ja":"We propose a real-time retrieval method for document images in various languages. In this method, queries are images of documents captured by a web-camera. The document images corresponding to the queries are retrieved from the document image database in real time. Since we have already proposed a document image retrieval method for English documents, the proposed method is an extension for retrieval of documents in various languages. In the previous English document image retrieval method, only centroids of word regions are used as feature points. Therefore it cannot be applied to some languages including Japanese and Chinese due to no separation between words and periodic arrangements of characters. In the proposed method, additional features are introduced to realize real-time retrieval for document images in various languages."}},{"key":"Sun2009_ICDAR","id":"1097","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Weihan Sun","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Detecting Printed and Handwritten Partial Copies of Line Drawings Detecting Printed and Handwritten Partial Copies of Line Drawings"},"journal":{"ja":"Proceedings of the 10th International Conference on Document Analysis and Recognition (ICDAR2009)","en":"Proceedings of the 10th International Conference on Document Analysis and Recognition (ICDAR2009)"},"pages":"341-345","year":2009,"month":7,"location":"Barcelona, Spain","review":"reviewed","fields":["pattern-recognition"],"doi":"10.1109/ICDAR.2009.149"},{"key":"Uchida2009_ICDAR","id":"1112","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Seiichi Uchida","en":""},{"ja":"Ryoji Hattori","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Conspicuous Character Patterns"},"journal":{"ja":"Proceedings of the 10th International Conference on Document Analysis and Recognition (ICDAR2009)"},"pages":"16-20","year":2009,"month":7,"location":"Barcelona, Spain","review":"reviewed","fields":["pattern-recognition"],"doi":"10.1109/ICDAR.2009.196","abstract":{"ja":"Detection of characters in scenery images is often a very\ndifficult problem. Although many researchers have tackled\nthis difficult problem and achieved a good performance, it\nis still difficult to suppress many false alarms andalthough\nmissings. This paper investigates a conspicuous character\npattern, which is a special pattern designed for easier\ndetection. In order to have an example of the conspicuous\ncharacter pattern, we select a character font with a\nlarger distance from a non-character pattern distribution\nand, simultaneously, with a smaller distance from a character\npattern distribution. Experimental results showed that\nthe character font selected by this method is actually more\nconspicuous (i.e., detected more easily) than other fonts."}},{"key":"Iwamura2009_CBDAR","id":"1095","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Tomohiko Tsuji","en":""},{"ja":"Akira Horimatsu","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Real-Time Recognition of Camera-Captured Characters in Complex Layouts"},"journal":{"ja":"Proceedings of the Third International Workshop on Camera-Based Document Analysis and Recognition (CBDAR2009)"},"pages":"53-60","year":2009,"month":7,"location":"Barcelona, Spain","review":"reviewed","fields":["document-image-analysis"],"abstract":{"ja":"In order to realize an application of camera-based character recognition, a character recognition technique which is (1) ready for real-time processing, (2) robust to perspective distortion, and (3) free from layout constraints, is required. In this paper, a camera-based character recognition method which satisfy the three requirements is proposed. As a result of a combination of simple but efficient techniques, the proposed method enables us to execute camera-based character recognition in real-time even on a laptop PC with a cheap web camera. Real-time processing is realized using a principle of geometric invariants in a different manner than usual. This drastically reduced computational cost and achieved fast recognition of around 200 to 250 characters per second on a server."}},{"key":"Kise2009_CBDAR","id":"1106","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Kazumasa Iwata","en":""},{"ja":"Tomohiro Nakai","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Shinichiro Omachi","en":""}],"title":{"ja":"Document-Level Positioning of a Pen Tip by Retrieval of Image Fragments"},"journal":{"ja":"Proceedings of the Third International Workshop on Camera-Based Document Analysis and Recognition (CBDAR2009)"},"pages":"61-68","year":2009,"month":7,"location":"Barcelona, Spain","review":"reviewed","fields":["document-image-analysis","document-image-retrieval"]},{"key":"Uchida2009_CBDARa","id":"1113","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Seiichi Uchida","en":""},{"ja":"Ryoji Hattori","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Selecting and Evaluating Conspicuous Character Patterns"},"journal":{"ja":"Proceedings of the Third International Workshop on Camera-Based Document Analysis and Recognition (CBDAR2009)"},"pages":"111-118","year":2009,"month":7,"location":"Barcelona, Spain","review":"reviewed","fields":["pattern-recognition"],"abstract":{"ja":"Character detection in scenery images is a very difficult\ntask. This paper describes a strategy of selecting character\npatterns for easier detection in scenery image. These\ncharacter patterns, called conspicuous character patterns,\nare selected from character font sets according to a criterion\nthat evaluates how the font has a larger distance from\na non-character pattern distribution and, simultaneously,\nwith a smaller distance from a general character pattern\ndistribution. Qualitative and quantitative evaluations have\nbeen made to observe how the selected fonts are more conspicuous\nthan other fonts."}},{"key":"Uchida2009_CBDARb","id":"1114","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Seiichi Uchida","en":""},{"ja":"Katsuhiro Itou","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"On a Possibility of Pen-Tip Camera for the Reconstruction of Handwritings"},"journal":{"ja":"Proceedings of the Third International Workshop on Camera-Based Document Analysis and Recognition (CBDAR2009)"},"pages":"119-126","year":2009,"month":7,"location":"Barcelona, Spain","review":"reviewed","fields":["document-image-analysis"],"abstract":{"ja":"Toward realization of “writing-life-log,” a camera-based\nhandwriting pattern acquisition system is proposed. The\ncamera is attached around the tip of a popular pen and captures\nframe images around the pen tip continuously. Our\nproblem is the reconstruction of the entire handwriting pattern\nby video-mosaicing of those frame images with perspective\nregistration of consecutive frames. A key idea is\nto use microscopic structure of paper surface, called paper\nfingerprint, for the registration. Specifically, perspective\ntransformation is estimated by using correspondence of\nSURF keypoints extracted on paper surface. Since the precise\nstructure can be captured stably as the SURF keypoints\nfrom the pen-tip camera, thus it is possible to expect accurate\nregistration of video frames."}},{"key":"Utsumi2009_TAPR","id":"1287","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"","en":"Yuzuko Utsumi"},{"ja":"","en":"Yoshio Iwai"},{"ja":"","en":"Masahiko Yachida"}],"title":{"en":"Face tracking based on 3D positional hypothesis"},"journal":{"en":"Proceedings on the 7th International Conference on Advances in Pattern Recognition"},"pages":"265-269","year":2009,"month":2,"location":"Kolkata, India","review":"reviewed","fields":[]},{"key":"Iwamura2009_XVI","id":"1050","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Ryo Niwa","en":""},{"ja":"Akira Horimatsu","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Shinichiro Omachi","en":""}],"title":{"ja":"Layout-free dewarping of planar document images"},"journal":{"ja":"Proc. Document Recognition and Retrieval XVI (Proc. SPIE Volume 7247)"},"presen_id":"7247-36","year":2009,"month":1,"location":"San Jose, California, USA","review":"reviewed","fields":["document-image-analysis"],"doi":"10.1117/12.806122","abstract":{"ja":"For user convenience, processing of document images captured by a\ndigital camera has been attracted much attention.\nHowever, most existing processing methods require\nan upright image such like captured by a scanner.\nTherefore, we have to cancel\nperspective distortion of a camera-captured image\nbefore processing.\nAlthough there are rectification methods of the distortion,\nmost of them work under certain assumptions on the layout;\nthe borders of a document are available, \ntextlines are in parallel, \na stereo camera or a video image is required and so on.\nIn this paper, we propose a layout-free rectification method\nwhich requires none of the above assumptions.\nWe confirm the effectiveness of the proposed method by experiments."}},{"key":"Nakai2009_XVI","id":"1051","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Tomohiro Nakai","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Camera-based document image mosaicing using LLAH"},"journal":{"ja":"Proc. Document Recognition and Retrieval XVI (Proc. SPIE Volume 7247)"},"presen_id":"7247-25","num_pages":"10","year":2009,"month":1,"location":"San Jose, California, USA","review":"reviewed","fields":["document-image-analysis","document-image-retrieval"],"doi":"10.1117/12.805949","abstract":{"ja":"In this paper we propose a mosaicing method of camera-captured document images. Since document images captured using digital cameras suffer from perspective distortion, their alignment is a difficult task for previous methods. In the proposed method, correspondences of feature points are calculated using an image retrieval method LLAH. Document images are aligned using a perspective transformation parameter estimated from the correspondences. Since LLAH is invariant to perspective distortion, feature points can be matched without compensation of perspective distortion. Experimental results show that document images captured by a digital camera can be stitched using the proposed method."}},{"key":"Sumioka2009","id":"1121","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"住岡 浩之","en":""},{"ja":"外山 託海","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"Chinese Restaurant Processを用いたVisual Wordの作成"},"journal":{"ja":"平成21年 電気関係学会関西支部連合大会 講演論文集"},"presen_id":"G11-8","year":2009,"month":11,"review":"not_reviewed","fields":["pattern-recognition","object-recognition"]},{"key":"Chikano2009","id":"1122","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"近野 恵","en":""},{"ja":"岩田 和将","en":""},{"ja":"黄瀬 浩一","en":""},{"ja":"岩村 雅一","en":""},{"ja":"内田 誠一","en":""},{"ja":"大町 真一郎","en":""}],"title":{"ja":"カメラペンシステムの検索精度向上のための一手法"},"journal":{"ja":"平成21年電気関係学会関西支部連合大会 講演論文集"},"presen_id":"G11-7","year":2009,"month":11,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"]},{"key":"Izuno2009_Conf","id":"1120","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"泉野 和弘","en":""},{"ja":"岩村 雅一","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"英文穴埋め問題のN-gram データを用いた解法"},"journal":{"ja":"平成21年電気関係学会関西支部連合大会 講演論文集"},"presen_id":"G11-18","year":2009,"month":11,"review":"not_reviewed","fields":["information-retrieval"],"awards":{"ja":["電子情報通信学会関西支部優秀論文発表賞受賞"]}},{"key":"Sasano2009_Conf","id":"1124","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"笹野 泰正","en":""},{"ja":"小島 篤博","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"人物行動に関する情報を付与したトポロジカルマップによる環境の表現"},"journal":{"ja":"平成21年電気関係学会関西支部連合大会 講演論文集"},"presen_id":"G13-19","year":2009,"month":11,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"計算機に室内環境を認識させる上で，人物行動に関する意味的な情報の表現も重要である．\nまた，2次元画像を用いたトポロジカルマップは処理が軽量であり，実用に適している．\n本研究では一つのアプローチとして，トポロジカルマップ上に認識しやすい観測地点を見つけることを考えた．\nまず，動作の対象となる物体領域の位置をトポロジカルマップ上に表現する．\nそして，それを観測できる地点の中からから認識のしやすい場所を選ぶ．"}},{"key":"Utsumi2009","id":"1297","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"内海ゆづ子","en":""},{"ja":"岩井儀雄","en":""},{"ja":"石黒浩","en":""}],"title":{"ja":"全方位視覚センサネットワークを用いた顔の追跡と認識"},"journal":{"ja":"第52 回自動制御連合講演会"},"presen_id":"F6-5","year":2009,"month":11,"review":"not_reviewed","fields":[]},{"key":"Inoue2009_IEICEPRMU_Experimental","id":"1125","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"井上 勝文","en":"Katsufumi Inoue"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"Bloomier Filterを用いた特定物体認識手法の性能に関する実験的検討","en":"Experimental Investigation of the Performance of a Specific Object Recognition Method with the Bloomier Filter"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"presen_id":"PRMU2009-109","pages":"115-120","year":2009,"month":11,"review":"not_reviewed","fields":["object-recognition","pattern-recognition"],"abstract":{"ja":"局所特徴量を表す特徴ベクトルを用いた物体認識手法では，膨大な数の特徴ベクトルを保存するために，莫大なメモリ容量が必要となる．この問題を解決するために，我々は空間効率の良いBloomier Filterを用いた手法を提案し，メモリ使用量の削減手法としての有効性を示している．しかし，特定物体認識手法の従来法と比べ，どの程度有効性があるのかという疑問が残っており，検討が必要である．そこで本稿では，従来提案されている``bag of features\"モデルに基づく手法と比較し，提案手法の有効性について検討する．具体的な結果として，1万個の平面物体を対象とした実験より，同じ認識率を得るために必要なメモリ使用量を1/7にできる．また，提案手法で用いている認識の評価基準と物体認識率との関係を実験的に解析した結果について述べる．","en":"Specific object recongnition based on local feature vectors requires a huge amount of memory to store all feature vectors. To solve this problem, we propose a memory reduction method with the Bloomier filter and show the its efficiency method for memory reduction in the previous paper. However, it is still unclear how efficient the proposed method is in comparison with conventional methods. In this report, we compare the efficiency of the proposed method with that of the method which is based on ``bag-of-features\" model. From the experiment with 10,000 planar objects, the proposed method allows us 1/7 required memory to obtain the same recognition rate compared with the conventional method. Additionally, in this report, we analyze experimentally the relationship between the recognition criteria with the proposed method and recognition accuracy."}},{"key":"Kise2009_IEICEPRMU","id":"1127","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"特定物体認識","en":"Specific Object Recognition"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"presen_id":"PRMU2009-104","pages":"79-87","year":2009,"month":11,"review":"not_reviewed","fields":["object-recognition"],"abstract":{"ja":"特定物体認識とは，画像内にある個別物体（物体のインスタンス）を\n認識する処理である．換言すれば，\n物体の「見え」から物体IDへの変換を意味する．\nこれが実現できれば，物体をWebのリンクアンカーとして\n用いることができ，Webを実世界に拡張することが可能となる．\n我々の周囲には無数の物体インスタンスがあるため，\n特定物体認識の実現は，規模との戦いであると言える．\n本稿では，局所特徴量を用いた特定物体認識を中心に，\n最近の動向について述べる．","en":"Specific object recognition is a process to recognize instances\nof objects captured as images. In other words, it indicates\nthe conversion from an appearance of an object to its ID.\nThis functionality allows us to use objects as link anchors \nof the web, and thus extend the cyber world to the real world.\nSince we are surrounded by an immense number of object instances,\nspecific object recognition, in its nature, is a challenge \nto the scalability. In this tutorial, I show recent advances in\nspecific object recognition with the special focus on\nmethods using local features."}},{"key":"Miki2009","id":"1128","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"三木 博史","en":"Hiroshi MIKI"},{"ja":"小島 篤博","en":"Atsuhiro KOJIMA"},{"ja":"黄瀬 浩一","en":"Koichi KISE"}],"title":{"ja":"人物動作のn-gram 表現に基づく物体認識","en":"Object Recognition Based on n-gram Representation of Human Action"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"presen_id":"PRMU2009-128","pages":"253-258","year":2009,"month":11,"review":"not_reviewed","fields":["object-recognition"],"abstract":{"ja":"近年の物体認識に関する研究において，物体に対する人物の動作を間接的に利用することで，外観的特徴\nだけでは判別が難しい物体の認識が試みられている．これまでの人物動作に基づく物体認識では，人物動作のカテゴ\nリを認識した後で，動作のカテゴリと物体とを関連付けるものがほとんどであった．本研究では，人物動作カテゴリ\nは分類せずに、物体に関わるプリミティブな人物の動きを直接利用することで物体を認識する手法を提案する．まず，\n動画像から人物の動作特徴を抽出し，これを記号化して一連の動作を記号列で表現する．これらの記号列からn-gram\nを抽出し，人物動作をn-gram の集合として扱う．そして，物体カテゴリに関わる動作のn-gram 集合を用い，入力動\n画像から得られた人物動作と対応させることで物体の認識を行う．本報告では，物体ごとに行った動作から物体を認\n識する実験を行い，提案手法の有効性を確認した．","en":"In the field of object recognition, usage of human actions is applied to it, in order to tackle with the\nrecongition of object that couldn’t be identified by only using their appearance. Most of the conventional research\nfor object recongition from actions, however, need an intermidiate classification of action category. In our research,\nprimitive motions of human are directly apply to object recongition without classifying action categories. First,\nfeatures of human motion are extracted from video images and encoded to symbols. Then, set of n-grams is obtained\nfrom the symbol sequence and registered to corresponding object category. At recognition phases, actions toward\nan object are also converted into n-grams which are to be compared with that registered in object categories. In\nthis paper, we performed experiments of object recognition using our methods and confirmed the validity of this."}},{"key":"Tada2009","id":"1126","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"多田 匡志","en":"Masashi Tada"},{"ja":"武藤 大志","en":"Tomoyuki Muto"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"近さの多段階表現に基づく近似最近傍探索","en":"Approximate Nearest Neighbor Search Based on a Multi-ValuedExpression of Closeness"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"presen_id":"PRMU2009-110","pages":"121-126","year":2009,"month":11,"fields":["nearest-neighbor-search"],"abstract":{"ja":"近似最近傍探索は，クエリと最も距離が近い点を探索する最近傍探索問題において，探索精度を犠牲にす\nることで計算時間，メモリ使用量を大幅に削減する手法である．本稿では，検索精度が同一であれば，ハッシュを用\nいた近似最近傍探索手法であるLSH やPCH より少ない計算時間とメモリ使用量で達成する手法を2 種類提案する．\n1 つ目の提案手法は，近さの情報を多値化して処理の効率化を図り，計算時間をLSH の約50%に削減した．2 つ目の\n提案手法は，既存の手法では不可欠だった距離計算の処理を省き，LSH に対してメモリ使用量を約50%，計算時間を\n約50%削減した．","en":"Approximate nearest neighbor search is a technique which greatly reduces processing time and required\namount of memory for nearest neighbor search. In this paper, we propose two methods which achieve the same\naccuracy, with less processing time and less required amount of memory compared to existing methods such as\nLSH and PCH. The first one achieved about 50% of processing time as compared to LSH by using multi-valued\ninformation for improving processing efficiency. The second one achieved about 50% of required amount of memory\nand about 50% of processing time as compared to LSH by eliminating distance calculation process which is requisite\nfor existing methods."}},{"key":"Tajima2009","id":"1117","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"田島 修司","en":""},{"ja":"内田 誠一","en":""},{"ja":"岩村 雅一","en":""},{"ja":"大町 真一郎","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"ペン先カメラ画像による手書きパターンの復元"},"journal":{"ja":"電気関係学会九州支部連合大会講演論文集"},"presen_id":"09-2A-12","year":2009,"month":9,"review":"not_reviewed","fields":["document-image-analysis"],"awards":{"ja":["電子情報通信学会九州支部長表彰 平成２１年度 連合大会講演奨励賞受賞"]}},{"key":"Yamashita2009","id":"1118","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"山下 健司","en":""},{"ja":"内田 誠一","en":""},{"ja":"岩村 雅一","en":""},{"ja":"大町 真一郎","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"検出容易な文字パターン生成に関する検討"},"journal":{"ja":"電気関係学会九州支部連合大会講演論文集"},"presen_id":"09-2A-13","year":2009,"month":9,"review":"not_reviewed","fields":["pattern-recognition","document-image-analysis"]},{"key":"Klinkigt2009_IEICEPRMU","id":"1152","type":"domestic_conf","language":"EN","show":"show","authors":[{"ja":"Martin Klinkigt","en":"Martin Klinkigt"},{"ja":"Koichi Kise","en":"Koichi Kise"},{"ja":"Heiko Maus","en":"Heiko Maus"},{"ja":"Andreas Dengel","en":"Andreas Dengel"}],"title":{"ja":"Object Detection in Images with Cluttered Background by Using Local Features and Their Configuration","en":"Object Detection in Images with Cluttered Background by Using Local Features and Their Configuration"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"Technical Report of IEICE"},"presen_id":"PRMU2009-65","pages":"75-80","year":2009,"month":8,"review":"not_reviewed","fields":["object-recognition"]},{"key":"Utsumi2009_MIRU","id":"1289","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"内海ゆづ子","en":""},{"ja":"岩井儀雄","en":""},{"ja":"石黒浩","en":""}],"title":{"ja":"全方位視覚センサネットワークを用いた顔の追跡と認識"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2008)"},"presen_id":"F6-5","pages":"1253-1260(IS2-58)","year":2009,"month":7,"review":"not_reviewed","fields":[]},{"key":"Inoue2009_MIRU","id":"1093","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"井上 勝文","en":"Katsufumi Inoue"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"Bloomier Filterを用いた特徴ベクトルの圧縮表現とその特定物体認識への応用","en":"Compressed Representation of Feature Vectors Using a Bloomier Filter and Its Application to Specific Object Recognition"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2009)論文集","en":"Proceedings of MIRU 2009"},"presen_id":"OS12-2","pages":"343-350","year":2009,"month":7,"review":"not_reviewed","fields":["object-recognition","pattern-recognition"],"abstract":{"ja":"特定物体を認識する手法に，局所特徴量を表す特徴ベクトルの最近傍探索を用いる手法がある．この手法では，特徴ベクトル間の距離を計算するために，特徴ベクトルそのものを記録する必要があるが，一般に物体認識に関与する特徴ベクトルの数は膨大であるため，莫大なメモリ容量が必要となる．この問題を解決するため，特徴ベクトルの距離を計算しない枠組みが必要となる．本論文では，特徴ベクトルの距離を計算しないメモリ使用量削減手法として，ハッシュ表などと比べてハルカに空間効率の良いBloomier Filterを用いる手法を提案する．また，平面および3次元特定物体の認識実験を通して，ハッシュ表を用いる手法と提案手法を比較し，有効性について検討する．","en":"Nearest neighbor search of feature vectors representing local features is often employed for specific object recognition. In such a method, it is required to store all feature vectors for distance calculation. The number of feature vectors is , in general, so large that a huge amount of memory is needed for their storage. A way to solve this problem is to skip the distance calculation for recognition. In this paper, we propose a method of object recognition without distance calculation. The characteristic point of the proposed method is to use a Bloomier filter, which is far memory efficinet than hash tables, for the storage of feature vectors. From experiments of planar and 3D specific object recognition, the proposed method is evaluated in comparison to a method with a hash table."}},{"key":"Iwamura2009_MIRU","id":"1094","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"辻 智彦","en":"Tomohiko Tsuji"},{"ja":"堀松 晃","en":"Akira Horimatsu"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"レイアウト非依存な実時間カメラベース文字認識","en":"Real-Time Camera-Based Character Recognition Free from Layout Constraints"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2009)論文集","en":"Proceedings of MIRU 2009"},"presen_id":"OS6-2","pages":"174-181","year":2009,"month":7,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"カメラを用いる文字認識の実用化を見据えると，(1) 実時間処理が可能である，(2) 射影歪みに頑健である，(3) 文字の配置に依らない認識が可能であるという3 つの要件を満たす実用的なカメラベース文字認識技術が求められる．本論文では，前述の3 要件を全て満たすカメラベース文字認識手法を提案する．提案手法は，単純だが効果的な処理を組み合わせることにより，安価なweb カメラと接続したノートパソコン上で実時間で動作する．実時間処理の実現のために，提案手法では幾何学的不変量の計算原理を通常とは異なる方法で用いた．その結果，認識処理に必要な計算量が大幅な削減され，サーバーを利用した場合には1 秒間に200〜250 文字程度の認識が可能になった．","en":"In order to realize an application of camera-based character recognition, a character recognition technique which is (1)ready for real-time processing, (2)robust to perspective distortion, and (3)free from layout constraints, is required. In this paper, a camera-based character recognition method which satisfy the three requirements is proposed. As a result of a combination of simple but efficient techniques, the proposed method enables us to execute camera-based character recognition in real-time even on a laptop PC with a cheap web camera. Real-time processing is realized using a principle of geometric invariants in a different manner than usual. This drastically reduced computational cost and achieved fast recognition of around 200 to 250 characters per second on a server."}},{"key":"Kise2009_MIRU","id":"1098","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":"Koichi KISE"},{"ja":"野口 和人","en":"Kazuto NOGUCHI"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"}],"title":{"ja":"参照特徴ベクトルの増大による特定物体認識の高速化と高精度化","en":"Improving Efficiency and Robustness of Specific Object Recognition by Increasing Reference Feature Vectors"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2009)論文集","en":"Proceedings of MIRU2009"},"presen_id":"OS12-1","pages":"335-342","year":2009,"month":7,"review":"not_reviewed","fields":["pattern-recognition","object-recognition"],"abstract":{"ja":"局所特徴量の照合による特定物体認識を考える．一般に，生成型学習を用いて，索引付けのための局所特徴量（参照特徴ベクトル）の数を増やせば，それだけ認識率が向上する．本論文では，認識率の向上だけではなく，処理時間も短縮可能であることを示す．参照特徴ベクトルの数が増えると処理時間が短縮される，という逆説的な効果は，識別器の多段階化による認識処理の早期終了によって得られる．これは，生成型学習によって照合に必要な探索空間を制限できるという効果による．1万画像を用いた認識実験の結果，6.6倍の参照特徴ベクトルを用いることで，処理時間を2/3, 認識率を12.2\\%改善できること，ならびに26億個の参照特徴ベクトルを用いて索引付けされた100万物体を，59ms/query, 90\\%の認識率で認識可能であることを示す．","en":"This paper concerns specific object recognition based on local features. In general, increasing the number of local features for indexing (reference feature vectors) by generative learning enables us to improve the recognition rate. In this paper, we show that generative learning is also effective for shorten the processing time. This paradoxical effect (i.e., shorter time with more reference feature  vectors) is achieved by cascading recognizers. Generated local features allow us to terminate the recognition process earlier. In other words, generative learning is effective to reduce the search space for finding nearest neighbors. From the experimental results using 10,000 reference images, 6.6 times reference feature vectors enabled us both to reduce the processing time to 2/3 from the original, and to improve the recognition rate by 12.2\\%. Another experiment with 1 million reference images indexed by 2.6 billion reference feature vectors yielded the recognition rate of 90\\% in 59ms/query."}},{"key":"Tsuji2009","id":"1096","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"辻 智彦","en":"Tomohiko TSUJI"},{"ja":"堀松 晃","en":"Akira HORIMATSU"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"},{"ja":"黄瀬 浩一","en":"Koichi KISE"}],"title":{"ja":"文字の並びをリンクアンカー化するWeb カメラベースインタフェース","en":"A Webcam-Based Interface Which Makes Arrangements of Characters toAnchor Texts"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2009)論文集","en":"Proceedings of MIRU2009"},"presen_id":"DS-3","pages":"1863-1864","year":2009,"month":7,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"ユーザーにとって快適なアプリケーションを考えたとき，インタフェースには即応性が要求される．我々は高速なカメラベース文字認識技術を提案しているが，これを利用すれば新たな対話型インタフェースになり得ると考えている．本デモではWeb カメラで撮影した環境中の文字を認識し，関連した情報を提供するシステムを提案する．","en":"It is necessary for an interface to have a rapid response in order to realize a comfortable application for the user. We have proposed a real-time camera-based character recognition method, and we think that there is a possibility to realize a new interactive interface with it. In this paper, we demonstrate a system which recognizes characters captured with a webcam and offers related information."}},{"key":"Uchida2009_MIRU","id":"1115","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"服部 亮史","en":"Ryoji Hattori"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"検出容易な文字パターン","en":"Designing Conspicious Character Patterns"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2009)論文集","en":"Proceedings of MIRU 2009"},"presen_id":"OS6-3","pages":"182-189","year":2009,"month":7,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"環境画像からの文字領域検出?切り出し処理\u0002 に関して，検出が容易となる文字パターンの合成について検討する．文字領域検出には，例えば文字パターンが一般的に持つ複雑性などの普遍的な特徴が用いられていた．しかし用いた特徴では検出できない例外的なパターンが存在したり，検出対象が存在する環境によっては検出が極めて困難になるなど問題も多い．本研究では，検出処理に用いる特徴を備えた文字パターンを合成するという視点からこの問題を捉える．具体的には，文字パターンの形状とある検出器による特徴抽出結果の関係を明らかにし，その検出器における処理に最適な文字パターン形状を定量的な評価により求めるというアプローチを用いる．この一例として，部分空間法を用いた文字量器検出に適した文字パターンを合成し，そのパターンの検出処理における有効性を確認した．","en":"Detection of characters in scenery images is often a very hard problem.\nAlthough many researchers have tackled this difficult problem and achieved a\ngood performance, it is still difficult to suppress many false alarms and/or\n missings. This paper investigates a conspicuous character pattern, which\nis a special pattern designed for easier detection. In order to have an\nexample of the conspicuous character pattern, we select a character font\nwith a larger distance from a non-character pattern distribution and,\nsimultaneously,\nwith a smaller distance from a character pattern distribution. Experimental\n results showed that the character font selected by this method is actually\n more conspicuous (i.e., detected more easily) than other fonts."}},{"key":"Iwamura2009_SSII","id":"1089","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"辻 智彦","en":"Tomohiko Tsuji"},{"ja":"堀松 晃","en":"Akira Horimatsu"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"カメラで撮像した文字画像の実時間認識システム","en":"A Real-Time Recognition System of Camera-Captured Characters"},"journal":{"ja":"第15回画像センシングシンポジウム(SSII09)講演論文集"},"presen_id":"IS3-28","year":2009,"month":6,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"本稿ではカメラを用いた文字認識技術の応用を見据えて，web カメラと接続したノートパソコン上で実時間で動作することができるカメラベース文字認識技術を提案する．提案手法は，実時間処理が可能であること，射影歪みに対応すること，様々なレイアウトの文書が認識可能であるという3 つの要件を同時に満たす実用的な技術である．"}},{"key":"Sun2009_SSII","id":"1090","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"孫 維瀚","en":"Weihan Sun"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"大規模データベースからの線画の部分的複製検出","en":"Partial Copy Detection of Line Drawings from a Large-Scale Database"},"journal":{"ja":"第15回画像センシングシンポジウム(SSII09)講演論文集"},"year":2009,"month":6,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"部分的複製とはオリジナル素材を部分的に切り出す\nことによって作成される複製である．不正な利用者は，\n発見を困難にするため，部分的複製を改変すると共に，\n他の素材の中に埋め込む形で使うことがある．本稿で\nは，このような場合にも対処可能な複製検出法を提案\nする．対象としては，漫画のような線画を取り上げる．\n線画の場合，改変の手法として手書き複製が考えられ\nるが，これが問題をより困難なものとしている．この\n問題に対処するため，本手法では，著作権保護対象の\n線画に対する局所特徴量の照合を考える．具体的には，\n局所領域の抽出手法としてMSER を，特徴量の記述子\nとしてHOG を用いる．また，11,603 ページの漫画の\nデータベースを用いた実験の結果から，提案手法が，印\n刷線画の部分的複製ばかりでなく，複雑な背景に埋め込\nまれた手書き線画に対しても有効性があることを示す．"}},{"key":"Inoue2009_IEICEPRMU_Memory","id":"1091","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"井上 勝文","en":"Katsufumi Inoue"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"特定物体認識におけるBloomier Filterを用いたメモリ削減","en":"Memory Reduction with Bloomier Filters for Specific Object Recognition"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"presen_id":"PRMU2009-56","pages":"101-106","year":2009,"month":6,"review":"not_reviewed","fields":["object-recognition","pattern-recognition"],"abstract":{"ja":"特徴ベクトルの最近傍探索を用いた特定物体認識では，特徴ベクトルの情報を記録するために莫大なメモリ量が必要となる．そこで，特徴ベクトル間の距離計算をしないことで，特定物体認識に必要なメモリの使用量を削減する枠組みを提案する．本手法の特徴は，ハッシュ表などと比べて空間効率の良いBloomier Filterを用いる点にある．また，平面および3次元特定物体の認識実験を通して，ハッシュ表を用いる手法と提案手法を比較し，有効性について検討する．","en":"Specific object recognition based on nearest neighbor search of feature vectors required a huge amount of memory to store all feature vectors for distance calculation. Tosolve this problem, we propose a memory reduction method for specific object recognition with a strategy of skipping the distance calculation of feature vectors. The proposed method is characterized by the use of Bloomier filters, which are far memory efficient than hash tables, for the storage of feature vectors. The proposed method is evaluated based on experiments of planar and 3D specific object recognition in comparison to a method with a hash table."}},{"key":"Muto2009_IEICEPRMU","id":"1092","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"武藤 大志","en":"Tomoyuki Muto"},{"ja":"多田 匡志","en":"Masashi Tada"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"隣接バケット探索を用いた近似最近傍探索手法の解析","en":"Efficient Approximate Nearest Neighbor Search Based on Accessing Neighboring Buckets"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Tecnical Report"},"presen_id":"PRMU2009-53","pages":"83-88","year":2009,"month":6,"fields":["nearest-neighbor-search"],"abstract":{"ja":"近似最近傍探索は，\nクエリと最も距離が近い点を探索する最近傍探索の\n計算量，メモリ使用量を大幅に削減する手法である．\n近似最近傍探索において，メモリ使用量ををさらに減少させる\nことが重要な課題である．\n本論文では，\n隣接バケットを参照する近似最近傍手法のモデル化を行い，\n近似最近傍探索手法の代表的な手法であるLSHよりも\nメモリ使用量を抑えて最近傍点を探索できることを\n実験と理論解析によって示す．","en":"Approximate nearest neighbor search is a technique which greatly\nreduces processing time and required amount of memory \nfor nearest neighbor search.\nFurther reduction of required amount of memory\nof the approximate nearest neighbor search \nis an important task.\nIn this paper, we model a method to\naccess neighboring buckets,\nand reveal the method requires less amount of memory than LSH\nby an experiment and analysis."}},{"key":"Kise2009_FDEI","id":"1054","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"野口 和人","en":"Kazuto Noguchi"}],"title":{"ja":"局所特徴量のハッシングに基づく大規模画像検索","en":"Large-Scale Image Retrieval Based on Hashing of Local Features"},"journal":{"ja":"データ工学と情報マネジメントに関するフォーラム論文集","en":"Proc. First Forum on Data Engineering and Information Management"},"presen_id":"D6-3","year":2009,"month":3,"review":"not_reviewed","fields":["document-image-retrieval","object-recognition"],"awards":{"ja":["DEIM2009 最優秀インタラクティブ賞受賞"],"en":["DEIM2009 Best Interactive Presentation Award Winner"]},"abstract":{"ja":"検索質問を画像とする画像検索は，バーコードの代替や著作権保護など，数多く\nのアプリケーションを持つ重要なタスクである．一般にカメラ付き携帯電話な\nどのデバイスで撮影される検索質問は，データベースの画像と同じものを写し\nていても，撮影条件の変動によりデータとしては全く異なるものとなるため，\n検索は容易ではない．\nこの問題に対処するため，近年，局所特徴量を用いた検索が注目されている．\n局所特徴量を用いた検索は，撮影条件にあまり影響を受けない検索を可能と\nする反面，索引付けに用いられる特徴量（特徴ベクトル）の数が膨大となるため，\n処理時間に関する問題が生じる．本稿では，この問題を解決する方策として，\n我々が取り組んでいるハッシュ表を用いた検索手法について述べる．\n検索対象としては，風景や人物を撮影した一般の画像に加え，\n文書画像という全く性質の異なるものを取り上げ，各々に有効な\n局所特徴量を示すとともに，両者の検索に共通する考え方を述べる．","en":"Image retrieval with query images is a task to find their \ncorresponding images (near duplicates) from the database. \nIt is an important task since it has many applications \nsuch as a replacement of barcodes and copyright protection of images. \nThe task of retrieval is not trivial \nsince query images are often far different \nfrom their corresponding images in the database.\nThis is due to various imaging conditions, e.g., occlusion,\nperspective distortion, different lighting conditions, \ndeforcus and motion blur.\nIn order to solve this problem, retrieval based on local featuers \nattracts considerable attention. In this method, images are\nindexed using feature vectors extracted from local regions of\nimages. Although they allow us the retrieval relatively robust\nto imaging conditions, they pose a problem of processing time \nsince the number of local features is huge.\nIn this report, we describe our new retrieval methods\nbased on hashing for solving the above problem.\nAs images in the database, we deal with ordinary images\nabout scnene, objects and people, etc., as well as document images\nthat are completely dissimilar to ordinary images.\nLocal featuers for each type of images and the common notion\nfor successful retrieval are shown."}},{"key":"Nouso2009_FDEI","id":"1077","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"能祖 正樹","en":"Masaki Nouso"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu iwamura"}],"title":{"ja":"Local Text Reuse Detection のLSHによる高速化","en":"Speeding up Local Text Reuse Detection by Locality-Sensitive Hashing"},"journal":{"ja":"データ工学と情報マネジメントに関するフォーラム論文集","en":"Proc. First Forum on Data Engineering and Information Management"},"presen_id":"D8-6","year":2009,"month":3,"review":"not_reviewed","fields":["information-retrieval"],"abstract":{"ja":"Local text resuse detectionとは，文書の部分的なコピーを発見する技術であり，文書の盗用検出に有効である．\nただし，この処理には，データベースの規模を大きくすると，処理時間が大幅にかかってしまうという問題がある．\nその原因は，大量のデータに対して最近傍探索を行うためである．\n最近傍探索にかかる時間を短縮する手法の一つに，近似を導入するものがある．\nそこで，本稿では近似最近傍探索の一手法であるLocality-Sensitive Hashing(LSH) とSpherical LSH を用いて，local text reuse detectionの処理を高速化する手法を提案する．","en":"Local text reuse detection is a technology to detect partial copies of documents and useful for anti-plagiarism of documents.\nA problem of methods for local text reuse detection is how to make the calculation of similarity of text parts efficiently, since it is generally based on nearest neighbor search whose cost is related to the size of the database.\nIn order to solve this problem, we propose a method of local text reuse detection with the help of approximate nearest neighbor search.\nMethods called Locality-Sensitive Hashing (LSH) and its variant called Spherical LSH are both employed to speed up the processing."}},{"key":"Hondo2009_IEICEPRMU","id":"1071","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"本道 貴行","en":"Takayuki HONDO"},{"ja":"黄瀬 浩一","en":"Koichi KISE"}],"title":{"ja":"特定物体認識のためのデータベース容量削減法の検討〜局所特徴量の量子化と取捨選択〜","en":"Inspection of Memory Reduction Methodsfor Specific Object Recognition—Approaches by quantization and Selection of Local Features"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"presen_id":"PRMU2008-265","year":2009,"month":3,"review":"not_reviewed","fields":["pattern-recognition"],"note":{"ja":"2009-CVIM-166 (26)"},"abstract":{"ja":"SIFT(Scale-Invariant Feature Transform) などの局所特徴量を用いて，大規模特定物体認識を行う場合には，データベースに保持しておく局所特徴量（特徴ベクトル）の数が増大する．そのため，メモリ容量の削減が課題\nとなる．本稿では，局所特徴量のベクトル量子化の手法と，取捨選択の方法によって，メモリ容量の削減を試みる．実験の結果，ベクトル量子化では，好ましい結果は得られなかった．一方，局所特徴量の取捨選択では，スケール耐性を犠牲にすることにはなるものの，入力画像のスケールがおおよそ決まっていれば，無削減のデータベースの1/10程\n度にしても，認識率はほとんど変化しないことが分かった．","en":"In case of conducting large-scale specific objects recognition using local features such as SIFT, the number of local features increases. Hence, reduction of the memory utilization is an important issue. In this report, we\nattempt memory reduction with two approaches; one is a method using vector quantization, and the other is one using selection of local features. From experimental results, we confirm that good results could not be acquired by vector quantization. On the other hand, if we employ feature selection as the method, the recognition rate hardly\ndecreases with a database whose size is 10% of its original."}},{"key":"Miyake2009_IEICEPRMU","id":"1074","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"三宅 弘志","en":"Hiroshi MIYAKE"},{"ja":"黄瀬 浩一","en":"Koichi KISE"}],"title":{"ja":"物体認識技術に基づく実物体からのリンク機能の実現","en":"Building links from real-world objects based on object recognition technologies"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"presen_id":"PRMU2008-271","year":2009,"month":3,"review":"not_reviewed","fields":[],"note":{"ja":"2009-CVIM-166 (32)"}},{"key":"Noguchi2009_IEICEPRMU","id":"1072","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"野口 和人","en":"Kazuto NOGUCHI"},{"ja":"氏原 慎弥","en":"Shinya UJIHARA"},{"ja":"黄瀬 浩一","en":"Koichi KISE"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"}],"title":{"ja":"カメラ付き携帯電話を入力デバイスとした大規模画像認識","en":"Large-Scale Image Recognition with a Camera Phone"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"presen_id":"PRMU2008-270","year":2009,"month":3,"review":"not_reviewed","fields":["pattern-recognition","object-recognition"],"awards":{"ja":["平成20（2008）年度 PRMU研究奨励賞受賞"],"en":["2008 PRMU Award Winner"]},"note":{"ja":"2009-CVIM-166 (31)","en":"2009-CVIM-166 (31)"}},{"key":"Hattori2009","id":"1060","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"服部 亮史","en":"Ryoji Hattori"},{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"検出容易な文字パターン生成に関する検討","en":"Conspicious Character Patterns"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"presen_id":"PRMU2008-209","year":2009,"month":2,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"情景画像中の文字領域を検出する処理に関して，検出が容易となる文字パターンの生成について検討する．従来の文字領域検出処理は，あらゆる文字パターンが共通して持つ普遍的な特徴を用いて文字領域と非文字領域を識別することで実現される．しかし用いた特徴では検出できない例外的なパターンが存在したり，検出対象が存在する環境によっては検出が極めて困難になるなどの問題も多い．本研究では，検出が容易となる特徴を備えた文字パターンを生成するという視点からこの問題を捉える．具体的には，検出特徴空間において非文字分布から最も離れた文字分布内の点を，検出が容易なパターンと定める．この考え方により生成されるパターンは，検出特徴，文字分布，非文字分布，パターン間の距離の定義に依存して異なってくる．その一例として部分空間法を応用したパターン生成手法を提案し，また提案手法により生成された文字の検出容易性を確認した．","en":"Characters in scene image are often hard to detect, i.e., not conspicuous. Thus, one of the main tasks for camera-based character recognition is the detection of characters in scene image. There are many past attempts for this difficult task. This paper investigates the essence of this task, that is, ``what is conspicuous character images?'' In order to have an example of the conspicuous character image, we use the relation between the subspace of non-character images and that of character images. Specifically, we try to select a image in the set of character images furthest from the subspace of non-character images."}},{"key":"Horimatsu2009_IEICEPRMU","id":"1058","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"堀松 晃","en":"Akira Horimatsu"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"点配置の一意性を利用した高速なアフィン不変形状認識","en":"Fast Affine Invariant Shape Recognition Using Uniqueness of　Arrangement of Points"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"presen_id":"PRMU2008-229","pages":"127-132","year":2009,"month":2,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"特徴点の配置を利用した物体認識法であるGeometric Hashing という手法が提案されている．この手法は，\n様々な座標系を考慮することで幾何学的変換にロバストである．Geometric Hashingをアフィン変換を受けた文字認識に適用することを考える．しかし，特徴点が多くなると座標系も多くなり，計算量が大きくなってしまう．そこで，座標系に制約を設けることで計算量を削減する方法を提案する．提案手法を用いた実験を行った結果，アフィン変換を受けた文字を高速に認識出来ることを確認した．","en":"Geometric hashing is an object recognition method using arrangements of feature points. This method\nis a robust retrieval method against geometric transformations by examining all of possible coordinate systems. We\nuse the geometric hashing to recognize characters undergone affine transformation. However, the more feature points are, the more coordinate systems are. Thus computational cost of recognizing characters is large. We propose a\nmethod to reduce the computational cost by restricting possible coordinate systems. By experiments, we confirmed that the proposed method can recognize characters faster."}},{"key":"Ito2009_IEICEPRMU","id":"1059","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"伊東 克啓","en":"Katsuhiro ITO"},{"ja":"内田 誠一","en":"Seiichi UCHIDA"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"},{"ja":"大町 真一郎","en":"Shinichiro OMACHI"},{"ja":"黄瀬 浩一","en":"Koichi KISE"}],"title":{"ja":"ペン先カメラ画像からの手書きパターンの抽出","en":"Recovering Handwritings via Pen-tip Camera"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"presen_id":"PRMU2008-211","year":2009,"month":2,"review":"not_reviewed","fields":["document-image-analysis"],"abstract":{"ja":"本研究では，日常の手書き内容を保存できるライティング・ライフ・ログの\n実現のための要素技術として，ペン先付近にカメラを搭載し，手書きパ\nターンをリアルタイムに取得・復元する方法を提案する．実際に手書き内容を得\nるためには，動画像の各フレームについてビデオモザイキングを行う必要\nがある．すなわち各フレーム画像について位置合わせをしながら統合し，\nパターン全体を復元する必要がある．本手法の特徴は，紙指紋と呼ばれる\n紙面の微細構造を積極的に利用する点にある．具体的には紙面よりSURF特\n徴点を抽出し，それに基づいてフレーム間の特徴点対応を求める．この対\n応関係を基にフレーム間の射影変換行列を求めて位置合わせを行う．これ\nを全フレームについて行うことで最終的に１枚の手書きパターン画像を生\n成する．幾つかの実例を通して，本手法の基本性能を検証した．","en":"Toward realization of ``writing-life-log'', a camera-based\n handwriting \npattern acquisition system is proposed. The camera is attached around\nthe tip of a popular pen. It captures frame images around the pen tip\ncontinuously. Our problem is video-mosaicing of those frame images by \nperspective registration of consecutive frames. A key idea is to use\nprecise structure of paper surface, called paper fingerprint, for the\n registration. Specifically, perspective transformation is estimated by\nusing correspondence of SURF feature points extracted on paper surface.\nSince the precise structure can be captured stably as the SURF feature \npoints from the pen-tip camera, thus it is possible to expect accurate\nregistration of video frames."}},{"key":"Iwamura2009_IEICEPRMU","id":"1052","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"印象的な研究を目指して —ユニバーサール・パターン・プロジェクトと大阪府立大学 知能メディア処理研究室における最近の研究とエピソード—","en":"Toward An Impressive Research —Recent Researches and Their Episodes of Universal Pattern Project and Intelligent Media Processing Laboratory, Osaka Prefecture University—"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"presen_id":"PRMU2008-217","year":2009,"month":2,"review":"not_reviewed","fields":["pattern-recognition","document-image-analysis"],"abstract":{"ja":"本稿では，筆者が携わっている文字認識・文書処理に関する研究の概要\nを述べるとともに，それらの研究の背景や\n研究が印象的になるように工夫したエピソードを紹介する．\nまた，文字・文書分野の現状や進むべき方向について筆者が個人的に\n感じていることについても言及する．","en":"In this report, I will introduce the overviews, backgrounds and\nepisodes of my recent researches related to\ncharacter recognition and document processing.\nI will also mention my personal opinion on the current status and\nrequirements of the research field."}},{"key":"Nakai2009_IEICEPRMU_Experimental","id":"1056","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"特徴点の局所的配置に基づく文書画像モザイキングの実験的検討","en":"Experimental Evaluation of Document Image Mosaicing Based on Local Arrangements of Feature Points"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"108","number":"432","presen_id":"PRMU2008-234","pages":"157-162","year":2009,"month":2,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"],"abstract":{"ja":"本稿では，デジタルカメラで撮影された，撮影範囲の異なる複数の文書画像をつなぎ合わせるモザイキング法を提案する．デジタルカメラで撮影された文書画像は任意の射影歪みを受けているため，その位置合わせは従来手法では困難なものであった．提案手法では，特徴点に基づく文書画像検索法であるLLAHを利用することで，射影歪みを受けた文書画像の特徴点の対応関係を得て，それに基づいて画像間の変換パラメータを推定して位置合わせを行う．LLAHは射影歪みに対して不変であるため，射影歪みの補正なしに特徴点の対応付けが可能である．実験により，デジタルカメラで撮影された文書画像のつなぎ合わせが可能であることが示された．","en":"In this paper we propose a mosaicing method of camera-captured document images. Since document images captured using digital cameras suffer from perspective distortion, their alignment is a difficult task for previous methods. In the proposed method, correspondences of feature points are calculated using an image retrieval method LLAH. Document images are aligned using a perspective transformation parameter estimated from the correspondences. Since LLAH is invariant to perspective distortion, feature points can be matched without compensation of perspective distortion. Experimental results show that document images captured by a digital camera can be stitched using the proposed method."}},{"key":"Nakai2009_IEICEPRMU_System","id":"1055","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"Webカメラを用いた多言語文書画像のリアルタイム検索システム","en":"A System of Real-Time Retrieval for Images of Documents in Various Languages using a Web Camera"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"108","number":"432","presen_id":"PRMU2008-227","pages":"115-120","year":2009,"month":2,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"],"abstract":{"ja":"本稿では，さまざまな言語で書かれた文書画像のリアルタイム検索法を提案する．これは，Webカメラで撮影された文書画像を検索質問とし，データベースから元となった文書画像をリアルタイムで検索して提示するものである．我々はすでに英語文書を対象とした文書画像検索法を提案しており，これは従来手法をさまざまな言語の文書に適用できるよう拡張したものである．従来手法である英語文書画像検索法では，画像処理によって得られる単語領域の重心を特徴点としていた．しかし，日本語や中国語を含むいくつかの言語では，言語の特性上識別性の高い特徴点を安定的に得ることが難しい．提案手法では，記述子の追加によってさまざまな言語の文書における高精度なリアルタイム文書画像検索を実現する．","en":"In this report, we propose a real-time image retrieval system using a web camera for documents in various languages. This system takes camera-captured document images as queries and find the corresponding document images from a database. This is an extension of our real-time image retrieval for English documents. In the English document image retrieval system, centroids of word regions are used as feature points to describe documents. These feature points are effective since English has clear between word gaps. On the other hand, in some languages including Japanese and Chinese, words are not separated. Therefore alternative feature points such as centroids of character regions are required. In the proposed method, this problem is solved by using the additional features extracted from character regions."}},{"key":"Niwa2009_IEICEPRMU","id":"1078","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"丹羽 亮","en":"Ryo Niwa"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"}],"title":{"ja":"レイアウトに依らない平面文書画像の射影歪み補正","en":"Dewarping of Planar Document Image without Layout Constraints"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"presen_id":"PRMU2008-233","year":2009,"month":2,"review":"not_reviewed","fields":["document-image-analysis"],"abstract":{"ja":"近年，デジタルカメラをスキャナの代わりに用いて文書画像処理を行う試みが盛んに研究されている．し\nかし，既存の文書画像処理技術の多くはスキャナで取得した文書画像を対象としているため，デジタルカメラで斜め\nから撮影した文書画像には適用できない．このような場合，文書画像に生じた射影歪みを補正し，スキャナで取得し\nたかのような真正面の文書画像を得る必要がある．既存の方法だと，文書の枠の有無や文字行の平行性あるいは，ス\nテレオカメラによる撮影やビデオカメラの使用などの制約がある．本稿では，文字の面積や変形に着目した自由なレ\nイアウトの文書へ適用できる歪み補正を提案し，実験により手法の有効性を確認する．","en":"For user convenience, processing of document images captured by a\ndigital camera has been attracted much attention.  However, most\nexisting processing methods require an upright image such like captured\nby a scanner. Therefore, we have to cancel perspective distortion of a\ncamera-captured image before processing. Although there are\nrectification methods of the distortion, most of them work under certain\nassumptions on the layout; the borders of a document are available,\ntextlines are in parallel, a stereo camera or a video image is required\nand so on. In this paper, we propose a layout-free rectification method\nwhich requires none of the above assumptions. We confirm the\neffectiveness of the proposed method by experiments."}},{"key":"Sun2009_PRMU","id":"1057","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"孫 維瀚","en":"Weihan Sun"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"局所特徴量の照合による線画の部分的複製検出","en":"Partial Copy Detection for Line Drawings by Local Feature Matching"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"The Institute of Electronics, Information and Communication Engineers"},"presen_id":"PRMU2008-228","pages":"121-126","year":2009,"month":2,"review":"not_reviewed","fields":["document-image-analysis"],"abstract":{"ja":"部分的複製とはオリジナル素材の一部を切り出すことによって作成される複製である．不正な利用者はこの手法を使って盗作を行うことが多い．その際には，部分的複製は他の素材の中に埋め込まれる形で使われる．加えて，部分的複製はそのままの形で埋め込まれるとは限らず，様々な処理によって改変されることも考えられる．その結果，不正利用の検出が困難となっている．そこで本稿では，このような場合にも対処可能な著作権保護の手法を提案する． 対象としては， 漫画のような線画を取り上げる． 線画の場合， 改変の手法として手書き複製が考えられるが，これが問題をより困難なものとしている．この問題に対処するため，本手法では，著作権保護対象の線画に対する局所特徴量の照合を考える． 具体的には， 局所領域の抽出手法としてMSER を， 特徴量の記述子としてHOG を用いる．実験により，提案手法は，印刷線画の部分的複製ばかりでなく，複雑な背景に埋め込まれた手書き線画に対しても有効性があることを示す．","en":"Abstract Partial copy is a kind of copy produced by cropping parts of original materials. Illegal users often use \t \nthis technique for plagiarizing　copyrighted materials. It is often the case that illegal partial copies are used by attaching them to other　materials. In addition, original parts are not necessarily copied intact but may be modified by various techniques, which make the detection quite　difficult. In this report, we propose a method of copyright protection applicable to partial copies aiming especially at the protection of line drawings such as cartoons. For the case of line drawings, handwriting can be a method of modification. In order to cope with handwritten partial　copies, we apply local feature matching with a database of copyrighted line drawings. We employ MSER as the region detector and HOG as the feature detector. Experimental results show that the proposed method not only performs good for detecting printed copies of line drawings, but also has effectiveness on the detection of handwritten ones, even if partial copies are embedded in complex backgrounds."}},{"key":"Matsumoto2009","id":"1290","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"松本雄大","en":""},{"ja":"内海ゆづ子","en":""},{"ja":"岩井儀雄","en":""}],"title":{"ja":"分枝限定法を用いた顔認識の効率化"},"presen_id":"PRMU2008-195","pages":"37-42","year":2009,"month":1,"review":"not_reviewed","fields":[]},{"key":"Hondo2009_MThesis","id":"1061","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"本道 貴行","en":"Takayuki Hondo"}],"title":{"ja":"大規模平面物体認識のための局所特徴量の比較とその選択法","en":"Comparison and Selection of Local Features for Large-scale Recognition of Planar Objects"},"year":2009,"month":3,"review":"not_reviewed","fields":["information-retrieval"]},{"key":"Horimatsu2009_MThesis","id":"1062","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"堀松 晃","en":"Akira Horimatsu"}],"title":{"ja":"点配置の一意性を利用した高速なアフィン不変文字認識","en":"Fast Affine Invariant Character Recognition UsingUniqueness of Arrangement of Points"},"year":2009,"month":3,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Miyake2009_MThesis","id":"1073","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"三宅 弘志","en":""}],"title":{"ja":"物体認識技術に基づく実物体からのリンク機能の実現"},"year":2009,"month":3,"review":"not_reviewed","fields":[]},{"key":"Niwa2009_MThesis","id":"1069","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"丹羽 亮","en":"Ryo Niwa"}],"title":{"ja":"レイアウトに依らない平面文書画像の射影歪み補正","en":"Dewarping of Planar Document Image without Layout Constraints"},"year":2009,"month":3,"review":"not_reviewed","fields":["document-image-analysis"]},{"key":"Noguchi2009_MThesis","id":"1070","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"野口 和人","en":"Kazuto Noguchi"}],"title":{"ja":"局所特徴量の近似最近傍探索による大規模特定物体認識","en":"Large-Scale Recognition of Specific Objects by Approximate Nearest Neighbor Search of Local Features"},"year":2009,"month":3,"review":"not_reviewed","fields":["object-recognition"]},{"key":"Nouso2009_MThesis","id":"1075","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"能祖 正樹","en":"Masaki Nouso"}],"title":{"ja":"ハッシュを用いたLocal Text Reuse Detection","en":"Local Text Reuse Detection with Hashing"},"year":2009,"month":3,"review":"not_reviewed","fields":["information-retrieval"]},{"key":"Izuno2009_BThesis","id":"1065","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"泉野 和弘","en":""}],"title":{"ja":"N-gramを用いた前置詞選択問題の解法の検討"},"year":2009,"month":3,"review":"not_reviewed","fields":["information-retrieval"]},{"key":"Muto2009_BThesis","id":"1063","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"武藤 大志","en":""}],"title":{"ja":"隣接バケット検索を用いた近似最近傍探索手法の解析"},"year":2009,"month":3,"review":"not_reviewed","fields":["information-retrieval"]},{"key":"Sasano2009_BThesis","id":"1068","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"笹野 泰正","en":""}],"title":{"ja":"トポロジカルマップにおける人物行動の観測地点の探索と評価"},"year":2009,"month":3,"review":"not_reviewed","fields":[]},{"key":"Toyama2009_BThesis","id":"1066","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"外山 託海","en":""}],"title":{"ja":"カテゴリの階層的表現を用いた物体認識"},"year":2009,"month":3,"review":"not_reviewed","fields":[]},{"key":"Ujihara2009","id":"1067","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"氏原慎弥","en":""}],"title":{"ja":"平面物体認識の大規模化と高速化"},"year":2009,"month":3,"review":"not_reviewed","fields":["object-recognition"]},{"key":"Kise2009","id":"1049","type":"survey","language":"JA","show":"show_except_datafile","authors":[{"ja":"黄瀬 浩一","en":""},{"ja":"岩村 雅一","en":""}],"title":{"ja":"物体認識における照合の高速化"},"journal":{"ja":"画像ラボ"},"pages":"65-71","year":2009,"month":1,"review":"not_reviewed","fields":[],"abstract":{"ja":"局所特徴量と呼ばれる特徴量を用いた物体認識では，他の特徴量と比べて高精度な認識が可能となる反面，画像から抽出される特徴量の数が膨大となるため，計算量が大きいという問題点を持つ．特に，局所特徴量を照合する処理において負荷が大きくなるため，その高速化が不可欠となる．本稿では，照合処理を特徴ベクトルの最近傍探索として捉え，近似を導入した近似最近傍探索によって，この問題が解決可能であることを示す．具体的には，以下の点について述べる．まず，物体認識のタスクを一般物体認識と特定物体認識の2種類に分類した後，双方において照合の高速化が重要であることを述べる．次に，高速化の手法として近似最近傍探索を取り上げ，具体的なアルゴリズムとしてANN（Approximate Nearest Neighbor）とLSH (Locality Sensitive Hashing)を紹介する．また，ANNによって実現した特定物体認識システムが，未知の画像と10万画像の照合を100ミリ秒程度で実行できることを示す．"}},{"key":"Ito2009_J","id":"1279","type":"other","language":"JA","show":"show","authors":[{"ja":"伊藤冬子","en":""},{"ja":"内海ゆづ子","en":""}],"title":{"ja":"学生フォーラム「オレ様より人様」平田圭二氏"},"journal":{"ja":"人工知能学会誌"},"volume":"24","number":"3","pages":"448-451","year":2009,"month":5,"review":"not_reviewed","fields":[]},{"key":"Utsumi2009_J","id":"1280","type":"other","language":"JA","show":"show","authors":[{"ja":"内海ゆづ子","en":""}],"title":{"ja":"学生インタビューと私"},"journal":{"ja":"人工知能学会誌"},"volume":"24","number":"1","pages":"38","year":2009,"month":1,"review":"not_reviewed","fields":[]},{"key":"Tanaka2008","id":"1022","type":"journal_paper","language":"JA","show":"show_except_datafile","authors":[{"ja":"田中 一弘","en":"Kazuhiro Tanaka"},{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"データ埋め込みペンに関する基礎的検討","en":"A Primary Study on a Data-Embedding Pen"},"journal":{"ja":"ヒューマンインタフェース学会論文誌","en":"Journal of Human Interface Society"},"volume":"10","number":"4","pages":"559-567","year":2008,"month":11,"review":"reviewed","fields":["pattern-recognition"],"doi":"10.11184/his.10.4_559"},{"key":"Uchida2008_IEICE","id":"1000","type":"letter","language":"JA","show":"show","authors":[{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"酒井 恵","en":"Megumi Sakai"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"事例に基づく文書画像の回転角推定","en":"Document Skew Estimation by Instance-Based Learning"},"journal":{"ja":"電子情報通信学会論文誌D","en":"Trans. IEICE"},"volume":"J91-D","number":"1","pages":"136-138","year":2008,"month":1,"review":"reviewed","fields":["pattern-recognition"],"url":"http://search.ieice.org/bin/summary.php?id=j91-d_1_136","abstract":{"ja":"各文字の回転変形に対する変量と不変量を事例として学習しておき，それらを利用することで文書画像の回転角を推定する方法を提案する．本手法は，文字単位で回転角を効率的に推定するため， 文字列が直線的かつ平行にレイアウトされているという仮定が不要であり，したがって様々なレイアウトの文書に利用可能である．","en":"A novel document skew (i.e., rotation) estimation technique proposed in this\n paper has two properties: First, it estimates the rotation angle at each\n connected component. The entire rotation angle is determined by voting all \n the estimated rotation angles. Second, it employs instance-based \n learning where the rotation angle is estimated by referring stored instances\n each of which consists of a rotation invariant and a rotation variant. \n The result of a skew estimation experiment on 55\n document images has shown that the skew angles of 54 document images \n were successfully estimated  with errors smaller than 2.0 degrees."}},{"key":"Miki2008","id":"1048","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Hiroshi Miki","en":""},{"ja":"Atsuhiro Kojima","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Environment Recognition Based on Human Actions Using Probability Networks"},"journal":{"ja":"2008 Second International Conference on Future Generation Communication and Networking"},"volume":"2","pages":"441-446","year":2008,"month":12,"review":"reviewed","fields":["pattern-recognition"],"abstract":{"ja":"To realize context aware applications for smart home environments, it is necessary to recognize function or usage of objects as well as categories of them. On conventional research for environment recognition in an indoor environment, most of previous methods are based on shape models. In this paper, we propose a method for recognizing objects focused on the relationship between human actions and functions of objects. Such relationship becomes obvious on human action patterns when he or she handles an object. To estimate object categories by using action patterns, we represent such relationship in Dynamic Bayesian Networks (DBNs). By learning human actions toward objects statistically, objects can be recognized. Finally, we performed experiments and confirmed that objects can berecognized from human actions without shape models."}},{"key":"Kise2008_ICPR","id":"1047","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Kazuto Noguchi","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Memory Efficient Recognition of Specific Objects with Local Features"},"journal":{"ja":"Proc. of the 19th International Conference of Pattern Recognition (ICPR2008)"},"presen_id":"WeAT3.1","num_pages":"4","year":2008,"month":12,"location":"Tampa, Florida, USA","review":"reviewed","fields":["object-recognition"],"doi":"10.1109/ICPR.2008.4761711"},{"key":"Inoue2008_KJPR","id":"1043","type":"intl_conf","language":"JA","show":"show","authors":[{"ja":"Katsufumi Inoue","en":""},{"ja":"Hiroshi Miyake","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"A Memory Reduction Method for 3D Object Recognition Based on Selection of Local Features"},"journal":{"ja":"Proceedings of Third Korea-Japan Joint Workshop on Pattern Recognition (KJPR2008)"},"pages":"7-8","year":2008,"month":11,"review":"reviewed","fields":["pattern-recognition"],"abstract":{"ja":"3D object recognition methods based on local features employ local features extracted from a large number of images for constructing object models and recognize 3D objects with these models. An important problem of these methods is requirement of huge memory space due to a large number of local features that constitute the object models. In this paper, we propose a simple method to reduce the required memory space by selecting local features in the object models. The proposed method is characterized by the mechanism of selection which employs estimates of the positive and negative effects of each local feature for object recognition. From experimental results for 30 objects, we achieved the recognition rate of 98.8% with about 10% of local features extracted from all images for models. Experimental results with COIL-100 (Columbia Object Image Library-100) show that 1/6 of the total local features allowed us the recognition rage of 96.8%."}},{"key":"Omachi2008","id":"1045","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Shinichiro Omachi","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Information Embedment with Cross Ratio of Areas for Accurate Camera-Based Character Recognition"},"journal":{"ja":"Proceedings of Third Korea-Japan Joint Workshop on Pattern Recognition (KJPR2008)"},"pages":"111-112","year":2008,"month":11,"location":"Seoul, Korea","review":"reviewed","fields":["pattern-recognition"]},{"key":"Tsuji2008_KJPR","id":"1044","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Tomohiko Tsuji","en":"Tomohiko Tsuji"},{"ja":"Masakazu Iwamura","en":"Masakazu Iwamura"},{"ja":"Koichi Kise","en":"Koichi Kise"}],"title":{"ja":"Generative Learning for Character Recognition of Uneven Lighting","en":"Generative Learning for Character Recognition of Uneven Lighting"},"journal":{"ja":"Proceedings of Third Korea-Japan Joint Workshop on Pattern Recognition (KJPR2008)","en":"Proceedings of Third Korea-Japan Joint Workshop on Pattern Recognition (KJPR2008)"},"pages":"105-106","year":2008,"month":11,"location":"Seoul, Korea","review":"reviewed","fields":["pattern-recognition"],"abstract":{"ja":"In order to improve recognition performance, we proposed a degradation model of uneven lighting for generative learning and experimentally confirm its effectiveness.","en":"In order to improve recognition performance, we proposed a degradation model of uneven lighting for generative learning and experimentally confirm its effectiveness."}},{"key":"Nakai2008_DASb","id":"1041","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Tomohiro Nakai","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Real-Time Retrieval for Images of Documents in Various Languages"},"journal":{"ja":"Handout of 8th IAPR International Workshop on Document Analysis Systems (DAS2008)"},"pages":"37","year":2008,"month":9,"location":"Nara, Japan","review":"reviewed","fields":["document-image-analysis","document-image-retrieval"],"doi":"10.1109/DAS.2008.22"},{"key":"Horimatsu2008_DAS","id":"1030","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Akira Horimatsu","en":""},{"ja":"Ryo Niwa","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Shinichiro Omachi","en":""}],"title":{"ja":"Affine Invariant Recognition of Characters by Progressive Pruning"},"journal":{"ja":"Proceedings of the 8th IAPR International Workshop on Document Analysis Systems (DAS2008)"},"pages":"237-244","year":2008,"month":9,"location":"Nara, Japan","review":"reviewed","fields":["pattern-recognition"],"doi":"10.1109/DAS.2008.88","abstract":{"ja":"There are many problems to realize camera-based character\nrecognition. One of the problems is that characters\nin scenes are often distorted by geometric transformations\nsuch as affine distortions. Although some methods that remove\nthe affine distortions have been proposed, they cannot\nremove a rotation transformation of a character. Thus\na skew angle of a character has to be determined by examining\nall the possible angles. However, this consumes\nquite a bit of time. In this paper, in order to reduce the\nprocessing time for an affine invariant recognition, we propose\na set of affine invariant features and a new recognition\nscheme called “progressive pruning.” The progressive pruning gradually prunes less feasible categories and skew angles using multiple classifiers. We confirmed the progressive pruning with the affine invariant features reduced the processing time at least less than half without decreasing the recognition rate."}},{"key":"Nakai2008_DASa","id":"1040","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Tomohiro Nakai","en":""},{"ja":"Kazumasa Iwata","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Accuracy Improvement and Objective Evaluation of Annotation Extraction from Printed Documents"},"journal":{"ja":"Proceedings of the 8th IAPR International Workshop on Document Analysis Systems (DAS2008)"},"pages":"329-336","year":2008,"month":9,"review":"reviewed","fields":["document-image-analysis","document-image-retrieval"]},{"key":"Uchida2008_DAS","id":"1042","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Seiichi Uchida","en":""},{"ja":"Megumi Sakai","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Skew Estimation by Instances"},"journal":{"ja":"Proceedings of the 8th IAPR International Workshop on Document Analysis Systems (DAS2008)"},"pages":"201-208","year":2008,"month":9,"location":"Nara, Japan","review":"reviewed","fields":["document-image-analysis"],"doi":"10.1109/DAS.2008.22","abstract":{"ja":"This paper proposes a novel skew estimation method by\ninstances. The instances to be learned (i.e., stored) are rotation\ninvariants and a rotation variant for each character\ncategory. Using the instances, it is possible to estimate\na skew angle of each individual character on a document.\nThis fact implies that the proposed method can estimate the\nskew angle of a document where characters do not form\nlong straight text lines. Thus, the proposed method will\nbe applicable to various documents such as signboard images\ncaptured by a camera. Experimental evaluation using\nsynthetic and real images revealed the expected robustness\nagainst various character layouts."}},{"key":"Iwamura2008_IEICEPRMU","id":"1046","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"吉村 宗浩","en":"Kazuhiro Yoshimura"},{"ja":"河野 修久","en":"Nobuhisa Kono"},{"ja":"楠 卓也","en":"Takuya Kusu"},{"ja":"平原 一帆","en":"Kazuho Hirahara"},{"ja":"高田 征吾","en":"Seigo Takada"},{"ja":"山崎 真也","en":"Shinya Yamazaki"},{"ja":"古畑 俊一郎","en":"Shuniichiro Fuuruhata"},{"ja":"佐藤 秀昭","en":"Hideaki Sato"},{"ja":"渡辺 大介","en":"Daisuke Watanabe"},{"ja":"正井 秀俊","en":"Hidetoshi Masai"}],"title":{"ja":"2008年PRMUアルゴリズムコンテスト　〜 「騎士(ナイト)を数えナイト 〜画像に含まれる特定物体の計数〜」 実施報告と受賞アルゴリズム紹介 〜","en":"PRMU Algorithm Contest 2008 -- ``Let's Count the Knights —Counting a Specific Object—,'' Summary Report and Prize Winning Algorithms --"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"presen_id":"PRMU2008-188","pages":"241-250","year":2008,"month":12,"review":"not_reviewed","fields":[],"abstract":{"ja":"電子情報通信学会 情報・システムソサイエティ パターン認識・メディア理解\n(PRMU)研究専門委員会が今年度実施したアルゴリズムコンテスト\n「\\ruby{騎士}{ナイト}を数えナイト 〜画像に含まれる特定物体の計数〜」\nの実施内容について報告すると共に，受賞者によるアルゴリズム紹介を掲載する．\n応募総数33件のアルゴリズムのうち，7件が入賞した．","en":"This article presents a summary report on the algorithm contest 2008\nentitled\n``Let's Count the Knights ---Count of a Specific Objects---,''\nwhich is sponsored by\nPRMU Technical Group of IEICE-ISS, with brief overviews of\nprize-winning algorithms provided by the winners.\n7 algorithms out of 33 have won the prizes."}},{"key":"Ujihara2008","id":"1053","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"氏原 慎弥","en":"Shinya Ujihara"},{"ja":"野口 和人","en":"Kazuto Noguchi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"大規模データベースを用いた画像認識の実験的検討","en":"Experimental Evaluation of Image RecognitionUsing a Large-Scale Database"},"journal":{"ja":"平成20年度情報処理学会関西支部 支部大会 講演論文集"},"pages":"173-176","year":2008,"month":10,"review":"not_reviewed","fields":["object-recognition"]},{"key":"Utsumi2008_MIRU","id":"1294","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"内海ゆづ子","en":""},{"ja":"岩井儀雄","en":""},{"ja":"谷内田正彦","en":""}],"title":{"ja":"３次元位置仮説に基づく顔の追跡と認識"},"journal":{"ja":"画像の認識・理解シンポジウム（MIRU2008)"},"pages":"1586-1591 (IS-5-32)","year":2008,"month":7,"review":"not_reviewed","fields":[]},{"key":"Hattori2008","id":"1037","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"服部 亮史","en":"Ryoji Hattori"},{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"環境画像からの検出が容易な文字パターンの合成に関する検討","en":"Conspicious Character Synthesis"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2008)論文集","en":"Proceedings of MIRU 2008"},"presen_id":"IS5-11","pages":"1456-1461","year":2008,"month":7,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"環境画像からの文字領域検出(切り出し処理) に関して，検出が容易となる文字パターンの合成について検\n討する．文字領域検出には，例えば文字パターンが一般的に持つ複雑性などの普遍的な特徴が用いられていた．しか\nし用いた特徴では検出できない例外的なパターンが存在したり，検出対象が存在する環境によっては検出が極めて困\n難になるなど問題も多い．本研究では，検出処理に用いる特徴を備えた文字パターンを合成するという視点からこの\n問題を捉える．具体的には，文字パターンの形状とある検出器による特徴抽出結果の関係を明らかにし，その検出器\nにおける処理に最適な文字パターン形状を定量的な評価により求めるというアプローチを用いる．この一例として，\n部分空間法を用いた文字領域検出に適した文字パターンを合成し，検出容易な文字パターンの性質を確認した．","en":"Characters in scene image are often hard to detect, i.e., not conspicuous. Thus, one of the main tasks\nfor camera-based character recognition is the detection of characters in scene image. There are many past attempts\nfor this difficult task. This paper investigates the essence of this task, that is, “what is conspicuous character\nimages?” In order to have an example of the conspicuous character image, we use the relation between the subspace\nof non-character images and that of character images. Specifically, we try to find a point in the character subspace\nfurthest from the non-character subspace."}},{"key":"Hondo2008","id":"1031","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"本道 貴行","en":"Takayuki HONDO"},{"ja":"黄瀬 浩一","en":"Koichi KISE"}],"title":{"ja":"大規模画像認識のための局所特徴量の性能比較","en":"Performance Evaluation of Local Features for Large-scale Image Recognition"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2008)論文集","en":"Proceedings of MIRU 2008"},"presen_id":"IS5-6","pages":"550","year":2008,"month":7,"review":"not_reviewed","fields":["document-image-analysis"],"abstract":{"ja":"画像を認識するための，様々な特徴点/領域の抽出手法や，局所記述子が提案されている．しかし，物体インスタンス（特定物体）を対象とした大規模画像認識において，どのような特徴点/領域を抽出し，どのような局所記述子を用いて，特徴量を記述することが最適なのかは十分明確であるとは言い難い．そこで本稿では，1万画像データベースを用いた大規模画像認識において，複数の特徴点/領域抽出手法，局所記述子を比較した結果を報告する．我々の実験では，特徴点/領域単独では，Hessian Affine 領域，局所記述子としてはGLOH やShape Context，また，特徴点と局所記述子を一括で扱うものの場合にはPCA-SIFT を用いることにより，最もよい結果が得られることが分かった．さらに，同一の記述子を用いても，特徴点/領域抽出の手法が異なれば，認識結果の誤りの傾向も異なり，特徴点/領域を組み合わせることにより精度が向上することが分かった．","en":"Various interest points/regions detectors and local descriptors have been proposed for image recognition.\nHowever, it is still unclear which interest points/regions detector or local descriptor gives the best performance for large-scale image recognition of object instances(specific objects). In this report, we present the results of comparison about performance of image recognition with several interest points/regions detectors and local descriptors using a 10,000 image-database. In our experiment, we confirmed that it was the best to employ Hessian-Affine Region as the interest points/regions detector and GLOH/Shape Context as the local descriptor. And PCA-SIFT, providing the functions of both the detector and the descriptor, gives the best performance as the combinational\nmethod. In addition, we confirmed that the tendency of misrecognition is different even if the same local descriptor is employed, in case of employing interest regions computed with another detector, furthermore, combination of detectors gives higher accuracy."}},{"key":"Horimatsu2008_MIRU","id":"1029","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"堀松 晃","en":"Akira HORIMATSU"},{"ja":"丹羽 亮","en":"Ryo NIWA"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"},{"ja":"黄瀬 浩一","en":"Koichi KISE"},{"ja":"内田 誠一","en":"Seiichi UCHIDA"},{"ja":"大町 真一郎","en":"Shinichiro OMACHI"}],"title":{"ja":"アフィン不変な文字認識手法とその高速化","en":"Affine Invariant Character Recognition and Its Speeding Up"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2008)論文集","en":"Proceedings of MIRU 2008"},"presen_id":"IS5-10","pages":"1450-1455","year":2008,"month":7,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"ディジタルカメラによる文字認識手法には課題が多く存在する．その課題の一つとして，幾何学変換によ\nり文字が歪むことが挙げられる．この課題に対し，歪みを正規化して認識する手法が存在する．しかし，この手法では\n文字の回転方向が推定できないため，方向を全て試すなどの処理が必要となり処理時間が大きくなる．そこで，本稿\nでは，アフィン不変な特徴量を用いて認識対象の字種を絞り込み，角度の推定を多段階で行うことで高速な認識を行\nうことを提案する．提案手法は，アフィン変換に対する頑健性を備え，認識時間を短縮できるという特徴を持つ．シ\nミュレーションと実際に撮影した文字に対して実験を行った結果，歪んだ文字に対し高速に認識することを確認する．"}},{"key":"Inoue2008_MIRU","id":"1026","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"井上 勝文","en":"Katsufumi Inoue"},{"ja":"三宅 弘志","en":"Hiroshi Miyake"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"局所記述子に基づく3次元物体認識のためのメモリ削減—局所記述子の取捨選択によるアプローチ—","en":"A Memory Reduction Method for 3D Object Recognition Based on Local Descriptors—An Approach by Selecting Local Descriptors—"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2008)論文集","en":"Proceedings of MIRU 2008"},"presen_id":"OS15-3","pages":"363-370","year":2008,"month":7,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"画像中に存在する3 次元物体を認識する手法として，局所記述子を用いるものがある．この手法は，大量の画像から抽出される局所記述子を用いてモデルを構築し，このモデルを用いて画像中の物体を認識する．この手法には，モデル構築に用いる局所記述子の数が莫大になり，これらを記憶しておくメモリ容量が膨大になるという問題点がある．本稿では，できるだけ少ない局所記述子で，物体を高精度で認識できるように，局所記述子を取捨選択し，メモリ容量を削減する手法を提案する．提案手法では，モデル構築用の画像を用いて物体認識シミュレーションを行い，認識率を落とす可能性の低い局所記述子をモデルに登録することで，高精度の物体認識を実現する低容量モデルを作成する．11 物体を用いた実験の結果，約100 分の1 に局所記述子の数を削減したモデルで，98%以上の認識率を\n得ることができた．またCOIL-100(Columbia Object Image Library-100)を用いた実験の結果，約6 分の1 に局所記述子の数を削減したモデルで，96%以上の認識率を得ることができた．","en":"3D object recognition methods based on local descriptors have been investigated. These methods em-\nploy local descriptors extracted from a lot of images to construct object models and recognize 3D objects with these\nmodels. An important problem of these methods is extensive memory requirement due to a large number of local\ndescriptors that constitute the models. In this report we propose a simple method to reduce the amount of memory\nby selecting local descriptors stored in the models. The proposed method is characterized by the mechanism of\nselection which employs an estimate of the ”contribution” of each local descriptor to object recognition. From\nexperimental results for 11 objects, we achieved the recognition rate over 98% with about 1% of local descriptors\nextracted from all images for models. Experimental results with COIL-100(Columbia Object Image Library-100),\nshow that 1/6 of the total local descriptors allowed us the recognition rate over 96%."}},{"key":"Iwamura2008_MIRU","id":"1023","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"古谷 嘉男","en":"Yoshio Furuya"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"内田 誠一","en":"Seiichi Uchida"}],"title":{"ja":"付加情報割当ての一般理論と誤認識率の解析","en":"A General Theory of Supplementary Information Assignment and  Analysis of Error Rates"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2008)論文集","en":"Proceedings of MIRU 2008"},"presen_id":"IS1-9","pages":"388-393","year":2008,"month":7,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"通常のパターン認識とは異なる，付加情報を用いるパターン認識という\n枠組みが提案されている．\nこれは通常のパターンと同時に，それを補助する情報(付加情報)を\nパターンとは別に用意し，それらを同時に用いることにより\n誤認識率を減少させる方式である．\n付加情報が誤らないという条件の下で誤認識率を最小にする条件が\nこれまでに導かれているが，\n実際には観測誤差が生じるために付加情報が誤らないという条件は\nいささか強過ぎる．\nそこで本研究では，付加情報に誤りがあるという条件において，\nどのような付加情報が誤認識率の減少に寄与するかについて考察する．\nさらにサンプルが正規分布に従い，各クラスの共分散行列が等しい場合について\n近似最適解を求め，誤認識率が減少することを検証する．","en":"Pattern recognition with supplementary information is a new\npattern recognition framework that determines an output class by\ncombining a feature vector extracted from the pattern and\nsupplementary information suggesting the true class.\nUnder the condition that supplementary information does not contain error,\na theory that reduces error rates\nhave been proposed.\nHowever, in the real world, we cannot observe any measure without error.\nThus, in this paper, we discuss how to reduce error rates\nusing the erroneous supplementary information, and\nconfirm the effect experimentally using artificial samples\nfollowing the normal distribution with a common covariance matrix."}},{"key":"Iwata2008_MIRU","id":"1038","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩田和将","en":"Kazumasa Iwata"},{"ja":"中居友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬浩一","en":"Koichi Kise"}],"title":{"ja":"文書からの書き込み抽出の精度向上と客観的評価","en":"Accuracy Improvement and Objective Evaluation of Annotation Extraction from Printed Documents"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2008)論文集"},"number":"IS4-20","pages":"1244-1251","year":2008,"month":7,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"]},{"key":"Miki2008_MIRU","id":"1034","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"三木 博史","en":"Hiroshi Miki"},{"ja":"小島 篤博","en":"Atsuhiro Kojima"},{"ja":"宮本 貴朗","en":"Takao Miyamoto"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"DBNを用いた人物の動作パターンに基づく物体認識","en":"Object Recognition Based on Human Actions by Using DBNs"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2008)論文集","en":"Meeting on Image Recognition and Understanding 2008"},"presen_id":"IS3-3","pages":"877-884","year":2008,"month":7,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"従来, 室内環境で自律的に動作するロボットのための環境認識の研究では,\n 主に物体の形状モデルに基づいた物体認識手法が提案されてきた. \n 本研究ではこういった物体の形状モデルを用いることなく, 人物の動作パター\n ンに着目した, 物体の確率的な推定手法を提案する. \n このような人物動作と物体の関連性は, 人物が特定の物体を扱う際に, 姿勢や手の\n 動きなどに特有のパターンとして表れる. この人物動作のパターンから物\n 体を推定するため, Dynamic Bayesian Networks (DBN)を用いて人物動作\n の特徴と物体との関係を確率的に表現する. さらに, このDBNに物体に対する人\n 物動作を与えて学習させることで, 物体の認識が可能となる.\n 最後に, 本手法の有効性を検証するため, 事前に物体の形状モデルを与え\n ることなく室内の物体を認識する実験を行った.","en":"On conventional research for enviornment recognition which is necessary \n for robots working autonomously in an indoor enviornment, most of \n previous methods are based on shape models.\n In this paper, we propose a method for object recognition focused on \n the relationship between human actions and objects.\n Such relationship becomes obvious on human action patterns when he or \n she handles an object.\n To estimate object categories by using action patterns, we represent such \n relationship probabilistically in Dynamic Bayesian Networks (DBN).\n By learning human actions toward objects statistically, objects can be recognized.\n Finally, we performed experiments and confirmed that objects can be \n recognized without shape models."}},{"key":"Nakai2008_MIRU","id":"1021","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"処理速度とメモリ効率の改善されたLLAHによるカメラベース文書画像検索","en":"Camera Based Document Image Retrieval with More Time and Memory Efficient LLAH"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2008)論文集","en":"Proceedings of MIRU 2008"},"presen_id":"IS4-21","pages":"1252-1259","year":2008,"month":7,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"],"abstract":{"ja":"本稿では，Locally Likely Arrangement Hashing (LLAH) を用いた文書画像検索の改善法を提案する．LLAHは高速かつ高精度でロバスト性が高いという長所をもつ一方で，必要メモリ量が多いという問題がある．また，リアルタイム文書画像検索などへ応用するためにはさらなる高速化が必要である．これらの問題点を解決するため，以下の2つの改良を施す．1つは重要性の低い特徴量をデータベースから削除し，データ構造を単純化することで実現される必要メモリ量の削減である．もう1つは，検索処理におけるアルゴリズムの改良による処理時間の短縮である．実験により，これらの改善によって必要メモリ量の80%，処理時間の60%が削減されることが確認された．","en":"In this paper, we propose improvements of our camera-based document image retrieval method with Locally Likely Arrangement Hashing (LLAH). While LLAH has high accuracy, efficiency and robustness, it requires a large amount of memory. It is also required to speed up the retrieval of LLAH for applications to real-time document image retrieval. For these purposes, we introduce the following two improvements. The first one is reduction of the required amount of memory by removing less important features for indexing from the database and simplifying structure of the database. The second improvement is to reduce exploring alternatives during the retrieval process. From the experimental results, we have confirmed that the proposed improvements realize reduction of the required amount of memory by about 80% and that of processing time by about 60%."}},{"key":"Niwa2008","id":"1032","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"丹羽 亮","en":"Ryo Niwa"},{"ja":"堀松 晃","en":"Akira Horimatsu"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"}],"title":{"ja":"変量と不変量と文字の正規化に基づく文書画像の射影歪み補正","en":"Rectification of Distorted Document Images Using Variants, Invariants and Normalization"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2008)論文集","en":"Proceedings of MIRU 2008"},"presen_id":"IS4-22","pages":"1260-1265","year":2008,"month":7,"review":"not_reviewed","fields":["document-image-analysis"],"abstract":{"ja":"近年，デジタルカメラをスキャナの代わりに用いて文書画像処理を行う試みが盛んに研究されている．し\nかし，既存の文書画像処理技術の多くはスキャナで取得した文書画像を対象としているため，デジタルカメラで斜め\nから撮影した文書画像には適用できない．このような場合，文書画像に生じた射影歪みを補正し，スキャナで取得し\nたかのような真正面の文書画像を得る必要がある．我々は既に変量と不変量に基づく射影歪みからアフィン歪みへの\n補正を提案している．これは，デジタルカメラの種類や文書の枠の有無，文字行の平行性を問題にしない補正へのア\nプローチである．本稿では手法の改良による精度向上，さらに書体情報を学習することで正対画像の復元の実現を実\n験結果とともに報告する．","en":"For user convenience, document image processing captured with a digital camera instead of a scanner\nhas been researched. However, existing methods of document image processing are not usable for a perspectively\ndistorted document image because most of them require a neat image captured by a scanner. Thus, we have to\nrectify perspective distortion of document images. For the purpose, we have proposed a layout free rectification\nmethod based on variants and invariants. In this paper, we reports two improvements on the method. The first one\nis to increase the stability of the method. The second one is to eliminate remaining affine distortions using prior\nknowledge. We confirm the effectiveness of the improvements by experiments."}},{"key":"Noguchi2008_MIRU_Experimental","id":"1024","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"野口 和人","en":"Kazuto NOGUCHI"},{"ja":"黄瀬 浩一","en":"Koichi KISE"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"}],"title":{"ja":"局所記述子に基づく物体認識のためのメモリ削減の実験的検討","en":"Experimental Evaluations of a Memory Reduction Methodfor Object Recognition Based on Local Descriptors"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2008)論文集","en":"Proceedings of MIRU 2008"},"presen_id":"OS10-3","pages":"251-258","year":2008,"month":7,"review":"not_reviewed","fields":["pattern-recognition","object-recognition"],"abstract":{"ja":"局所記述子を用いた物体認識において，認識率，処理時間，メモリ量について良いバランスを実現することは重要な問題である．従来，局所記述子をベクトル量子化してvisual word を得，それを認識に用いる手法が多数提案されている．インスタンス(特定物体) を対象とした認識では，数多くのvisual word を用いる手法は高い認識率が得られるものの，処理時間やメモリ量に問題が生じる．一方，局所記述子を量子化せずそのまま用いる手法であっても，近似最近傍探索によって処理が大幅に短縮できることが知られている．本稿では，ベクトル量子化を行わない場合でも，メモリ量の問題を解決する手法として，特徴ベクトルの各次元を記録するビット数を削減する手法を提案する．10 万画像のモデルを用いて認識実験を行った結果，各次元を2 ビットで表現したとき，ビット数削減前(16 ビット) と比べて，認識率は少し下がるものの，処理時間は殆ど変化しないことが分かった．","en":"For object recognition based on local descriptors, balancing the recognition rate, processing time and memory requirement is an important issue. In conventional methods, local descriptors are vector-quantized to “visual words” for describing images. Although many visual words allow us to improve the recognition rate for\nrecognizing instances (specific objects), they also pose a problem of processing time and memory requirement. On the other hand, even if a recognition method employs local descriptors without vector quantization, approximate nearest neighbor search enables us to cut the processing time drastically. In this report, we propose a method of memory reduction applicable even to methods without vector quantization. The proposed method reduces the memory requirement by limiting the number of bits for the record of each dimension of feature vectors. From experimental results on 100,000 images, it has been confirmed that a representation with 2 bit / dimension subtly decreases the recognition rate, but hardly changes the processing time."}},{"key":"Noguchi2008_MIRU_Relation","id":"1025","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"野口 和人","en":"Kazuto NOGUCHI"},{"ja":"黄瀬 浩一","en":"Koichi KISE"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"}],"title":{"ja":"物体の認識率と局所記述子の照合精度の関係","en":"Relation Between the Object Recognition Rate and the Matching Accuracy of Local Descriptors"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2008)論文集","en":"Proceedings of MIRU 2008"},"presen_id":"IS4-4","pages":"504","year":2008,"month":7,"review":"not_reviewed","fields":["pattern-recognition","object-recognition"],"abstract":{"ja":"局所記述子を用いた物体認識の最も単純な手法は，物体モデルに保存された局所記述子と未知の画像から抽出した局所記述子を最近傍探索によって照合し，投票するものである．一般に，局所記述子の数は膨大であるため，このアプローチの鍵は，いかに照合の計算コストを削減するかにある．近似最近傍探索の技術を用いると，照合に必要なコストが劇的に削減されるが，どの程度の照合精度があれば，高精度な物体認識が可能なのかという疑問点が残る．本稿では，この点について次の２つの成果を示す．まず，局所記述子の照合精度はそれほど高くなくてもよいことを実験的に示す．例えば，10 万物体に対して，97%の認識率は14%の照合精度で得られる．もう一つは，単純な2項分布を用いた，投票による認識のモデルを提案し，実験結果と良く合うことを示す．","en":"A simple method of object recognition with local descriptors is voting by matching local descriptors with nearest\nneighbor (NN) search. Because the number of local descriptors is so large, a key technology for this approach is to reduce\nthe computational cost of matching. It is known that the cost of NN search can be drastically cut by approximation, though\nit poses another question of how accurate the matching of local descriptors should be for accurate object recognition. The\nmajor contribution of this report is twofold. First, we experimentally show that it is not necessary for matching to be accurate.\nFor example, an object recognition rate of 97 % with 100,000 object models is achieved by the matching accuracy of only\n14%. Secondly, we propose a model of recognition by voting based on a simple binomial distribution, which agrees with\nexperimental results."}},{"key":"Umemoto2008","id":"1036","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"梅本 敏孝","en":"Toshitaka Umemoto"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"SIFT における特徴点候補検出の複素一次系による高速化","en":"Speeding up the Detection of Scale-Space Extrema in SIFT Based on the Complex First Order System"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2008)論文集","en":"Proceedings of MIRU 2008"},"presen_id":"IS2-15","pages":"682-687","year":2008,"month":7,"review":"not_reviewed","fields":[],"abstract":{"ja":"SIFT などの局所記述子の最近傍探索によって物体認識を行う場合には，画像から特徴量を抽出するのに必要な処理コストを最小化することが重要である．その特徴量を抽出する処理は，スケールとキーポイント検出，キーポイントのローカライズ，オリエンテーションの算出と特徴量の記述から構成されている．本論文では，これらの処理の中でスケールとキーポイント検出に必要な処理コストを削減する方法を提案する．具体的には，スケールを検索するための画像は，入力画像データとガウス関数の畳み込みによって作られる．そこで，その処理に必要な乗算回数\nを青島によって提案された複素1次系を用いることで削減する．処理時間の計測とANN を用いた画像検索実験から，提案した方法は認識率でガウス関数を用いた畳み込みを用いた場合と同程度の認識率を得ることができた．さらに，計算時間は3分の1 となった．","en":"For object recognition based on nearest neighbor search of local descriptors such as SIFT, it is important\nto minimize the cost of extracting image features. The processes of extracting image features are composed of\nscale-space extrema detection, keypoint localization, orientation assignment and keypoint descriptor.In this paper,\nwe propose a new method of efficient scale-space extrema detection. The scale space of an image is defined as\na function, that is produced from the convolution of a variable-scale Gaussian with an input image. Number in\nwhich multiplication is executed is reduced by complex first order system proposed by Aoshima. From experimental\nresults with computation time and recognition of image with ANN, we have confirmed that the proposed method is\ncapable of achieving a recognition rate as same as the original method, and 1/3 of the computation time with the\noriginal method."}},{"key":"Utsumi2008_CVIM","id":"1291","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"内海ゆづ子","en":""},{"ja":"岩井儀雄","en":""},{"ja":"谷内田正彦","en":""}],"title":{"ja":"３次元位置仮説による顔の追跡と認識"},"presen_id":"CVIM, No. 162-77","pages":"495-500","year":2008,"month":3,"review":"not_reviewed","fields":[]},{"key":"Sezaki2008","id":"1015","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"瀬崎 直人","en":"Naoto Sezaki"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"タグの共起と類似画像を利用したタグ付け支援システム","en":"Tagging System Using Co-occurrence of Tags and Similar Images"},"journal":{"ja":"電子情報通信学会 第19回データ工学ワークショップ論文集"},"year":2008,"month":3,"review":"not_reviewed","fields":[],"url":"http://www-nishio.ist.osaka-u.ac.jp/dews2008/proceedings/program.html","note":{"ja":"DEWS2008 E10-5"},"abstract":{"ja":"本稿では，タグの共起と類似画像を利用して，画像に対して有効なタグをユーザに提示するシステムを提\n案する．十分にタグ付けされたWeb 画像のデータベースから，タグの共起に加えて類似画像の関連語も抽出すること\nで，多様で精度の高い関連語をユーザに提示する．また，ユーザが新たに付与したタグを入力のタグに含め，関連語\nの再検索を行うことで，さらに新しい関連語を抽出することができる．16002 枚の画像を対象にして，タグの共起の\nみを用いる手法，また類似画像のみを用いる手法と比較実験を行った結果，提案手法はより多くの有効な関連語を抽\n出できることがわかった．また，これらの有効なタグを入力タグに追加し，関連語の再検索を行うことにより，新た\nに有効な関連語を抽出できることも明らかになった．","en":"In this report，we propose a method for recommending tags of images．The proposed method presents\nto the user various tags with high precision by taking into account both the co-occurrence of tags and similar images．\nAdditional search with the user feedback enables us to find some new tags relevant to the image of interest．In\norder to test the proposed method，we compare it with a method based only on the co-occurrence of tags，as well\nas a method based only on the similarity of images．From the experimental results using 16002 images，we have\nconfirmed that the proposed method is capable of finding more tags as compared to the methods for comparison．\nWe have also confirmed that the additional search is effective to find additional tags．"}},{"key":"Yokota2008","id":"1014","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"横田 哲","en":"Satoshi YOKOTA"},{"ja":"黄瀬 浩一","en":"Koichi KISE"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"},{"ja":"汐崎 陽","en":"Akira SHIOZAKI"}],"title":{"ja":"局所記述子に基づく幾何学的変換にロバストな画像著作権保護"},"journal":{"ja":"電子情報通信学会 第19回データ工学ワークショップ論文集"},"year":2008,"month":3,"review":"not_reviewed","fields":[],"url":"http://www-nishio.ist.osaka-u.ac.jp/dews2008/proceedings/program.html","note":{"ja":"DEWS2008 C1-5"},"abstract":{"ja":"近年，インターネット上でコンテンツの不正コピーによる著作権の侵害が社会的な問題となっている．著\n作権侵害の防止策としては，電子透かしというデジタルコンテンツに著作権情報を埋め込む技術がある．しかし画像\nに関して言えば，画像を改竄し，埋め込まれた情報を取り出せなくしてしまう攻撃(Stirmark 攻撃など) がある．特\nに，近年，幾何学的変換への耐性が問題となっている．本稿では，この問題を解決するため，画像の局所記述子を用\nいた新しい電子透かし法を提案する．画像の局所記述子は，幾何学的変換に影響を受けにくいという特性があるため，\n従来の問題点を解決できる可能性がある．本手法の特徴は，ノンブラインド型の電子透かし法であるため，局所記述\n子で定められる，画像の局所領域に順序をつけて，電子透かし情報を埋め込める点にある．これにより，透かし情報\nの容量を増加させることが可能となる．実験では，Stirmark 攻撃を受けた画像から，正しく局所領域を連結して透か\nし情報が読み出せるかどうかを検証する．","en":"In recent years, illegal copies on the Internet have become a serious social problem. As a method of\ncopyright protection, digital watermarking is a well-known mean which embeds copyright information to digital\ncontents to be protected. However, for the case of digital still images, for example, there exist image falsification\nattacks (e.g., Stirmark ) which change images to make it difficult to read the embedded copyright information. In\nparticular, robustness against geometric transformation becomes an important issue. In this paper, we propose a\nmethod for solving this problem by using “local descriptors” of images. Local descriptors have a property insensitive\nto geometric transformation, they have a potential to solve the problem. The characteristic feature of the proposed\nmethod is that it can sequence local regions defined by local descriptors for embedding information. This indicates\nthat the amount of information increases as compared to independent embedment on each local region. From the\nexperimental results, we show that the sequence of local regions can be correctly read and embedded information\ncan be successfully decoded from most of images that undergo the Stirmark attack."}},{"key":"Sakai2008","id":"1009","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"酒井 恵","en":"Megumi Sakai"},{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"文書画像の事例ベース傾き補正の改良","en":"An Improvement of Instance-Based Skew Estimation"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"number":"491","presen_id":"PRMU2007-214","pages":"1-6","year":2008,"month":2,"review":"not_reviewed","fields":["document-image-analysis"],"abstract":{"ja":"文書画像の回転(スキュー)を補正するため，事例ベース傾き補正法を提案している．事例を用いることにより，文書の直進性を仮定せずに回転角を\n推定できる．また，事例として回転変量と不変量を用いることにより効率的に回転角を推定できる．変量と不変量はどのようなものでも良いが，前回の報告で用いていた不変量はノイズに弱いという問題があった．そこで本報告では，回転角推定の精度向上を目的として，不変量の多次元化について述べる．計算機内実験において文書画像55サンプルを用いて，簡単な傾き推定実験を行ったところ，48サンプルにおいて誤差1度以下，全てのサンプルにおいて誤差2度以下という精度を得た．また，スキャナ取得画像に対しても評価実験を行った所，目視ではあるが入力10画像中9画像に対して正しく補正できた．","en":"The purpose of this report is to improve an instance-based deskewing\ntechnique which is free from the conventional assumption that text lines are\nstraight and parallel. The instances describe the relation\namong the skew angle, a skew variant, and a skew invariant in an compact\nmanner. A main idea of the improvement is to increase the number of the skew\ninvariants for more stable estimation of the skew angle. An experimental\nresult on 55 document images showed that their skew angles were successfully\nestimated with errors smaller than 2.0 degrees."}},{"key":"Nakai2008_PhD","id":"1308","type":"doctoral_thesis","language":"EN","show":"show_except_datafile","authors":[{"ja":"Tomohiro Nakai","en":""}],"title":{"ja":"Study on Hash-Based Document Image Retrieval and Its Application to Information Recycle"},"year":2008,"month":1,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"]},{"key":"Furuya2008","id":"1004","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"古谷 嘉男","en":""}],"title":{"ja":"付加情報の誤りと誤認識率の関係"},"year":2008,"month":3,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"satoshi2008","id":"1006","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"横田 哲","en":""}],"title":{"ja":"局所記述子を用いた幾何学的変換にロバストな画像著作権保護"},"year":2008,"month":3,"review":"not_reviewed","fields":["information-retrieval","pattern-recognition"]},{"key":"Sezaki2008_MThesis","id":"1005","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"瀬崎 直人","en":""}],"title":{"ja":"タグの共起と画像の類似性を利用したタグ付け支援システム"},"year":2008,"month":3,"review":"not_reviewed","fields":["information-retrieval"]},{"key":"Tanimoto2008","id":"1007","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"谷本 真吾","en":""}],"title":{"ja":"HTML文書に含まれる箇条書きの項目間の順序特徴に基づく分類"},"year":2008,"month":3,"review":"not_reviewed","fields":["information-retrieval"]},{"key":"Inoue2008_BThesis","id":"1018","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"井上 勝文","en":""}],"title":{"ja":"局所特徴量を用いた3次元物体認識のためのメモリ容量の削減"},"year":2008,"month":3,"review":"not_reviewed","fields":["pattern-recognition","information-retrieval"]},{"key":"Ito2008","id":"1013","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"伊藤 恭介","en":""}],"title":{"ja":"可変長パッセージにおける索引語の共起を考慮した文書検索"},"year":2008,"month":3,"review":"not_reviewed","fields":["information-retrieval"]},{"key":"Iwata2008_BThesis","id":"1019","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"岩田 和将","en":""}],"title":{"ja":"文書からの書き込み抽出の精度向上と客観的評価"},"year":2008,"month":3,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"]},{"key":"Sumioka2008","id":"1008","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"住岡 浩之","en":"Hiroyuki Sumioka"}],"title":{"ja":"大量データを学習に用いた語義曖昧性解消","en":"Word Sense Disambiguation by Large Training Corpus"},"year":2008,"month":3,"review":"not_reviewed","fields":["information-retrieval"]},{"key":"Tada2008","id":"1012","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"多田 匡志","en":""}],"title":{"ja":"ハッシュを用いた近似最近傍探索と計算量削減"},"year":2008,"month":3,"review":"not_reviewed","fields":["information-retrieval"]},{"key":"Tsuji2008_BThesis","id":"1010","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"辻 智彦","en":""}],"title":{"ja":"不均一照明に頑健な文字認識のための生成型学習"},"year":2008,"month":3,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Kise2008_IPSJ_Building","id":"1027","type":"survey","language":"JA","show":"show_except_datafile","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"3日で作る高速特定物体認識システム","en":"Building an Efficient System of Specific Object Recognition in Three Days"},"journal":{"ja":"情報処理","en":"IPSJ Magazine"},"volume":"49","number":"9","pages":"1082-1089","year":2008,"month":9,"review":"not_reviewed","fields":[]},{"key":"Kise2008_IPSJ_Years","id":"1028","type":"survey","language":"JA","show":"show_except_datafile","authors":[{"ja":"Andreas Dengel (翻訳：黄瀬 浩一)","en":"Andreas Dengel (Translated by Koichi Kise)"}],"title":{"ja":"ドイツ人工知能研究センター(DFKI)の20年—成功への道筋とそれを可能とした人々—","en":"20 Years of \"Deutsches Forschungszentrum für Künstliche Intelligenz (DFKI)\"—Chapters and Actors of a Success Story—"},"journal":{"ja":"情報処理","en":"IPSJ Magazine"},"volume":"49","number":"7","pages":"810-817","year":2008,"month":7,"review":"not_reviewed","fields":[]},{"key":"Uchida2008","id":"1020","type":"survey","language":"JA","show":"show_except_datafile","authors":[{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"酒井 恵","en":"Megumi Sakai"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"事例に基づく文書画像の回転角推定法"},"journal":{"ja":"画像ラボ"},"pages":"5-10","year":2008,"month":7,"publisher":"日本工業出版","review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"画像中の文書や文字列を認識する際，クラスへの識別処理に先立ち，二値化など様々な前処理が画像に施される．その一つに回転（スキュー）補正がある．これは文字行を正確に水平にするための処理であり，これを怠ると文字行の切り出しに失敗し，最悪の場合，全く認識不能となる．文字行の回転角は事前には不明であるため，補正の際には回転角を推定する必要がある．これまでにも様々な方法が提案されているが，いずれも文字行の直進性を仮定している．従って，環境中の文字・文書認識の対象となるであろう，ポスターや看板など短い文字列が自由にレイアウトされたような文書画像には，これら従来法は適さない．本稿では，この文字行直進仮定が不要な回転補正法を紹介する．"}},{"key":"Kise2008_OR","id":"1079","type":"other","language":"JA","show":"show_except_abstract","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"物体認識における照合の高速化"},"journal":{"ja":"第14回画像センシングシンポジウムダイジェスト集","en":"Symposium on Sensing via Image Information"},"presen_id":"OR4-03","pages":"OR4-03-1 - OR4-03-4","year":2008,"month":6,"review":"not_reviewed","fields":[]},{"key":"Utsumi2008_J","id":"1281","type":"other","language":"JA","show":"show","authors":[{"ja":"内海ゆづ子","en":""},{"ja":"伊藤冬子","en":""}],"title":{"ja":"学生フォーラム「自分の能力に境界線を引くな」石黒浩氏"},"journal":{"ja":"人工知能学会誌"},"volume":"23","number":"3","pages":"429-432","year":2008,"month":5,"review":"not_reviewed","fields":[]},{"key":"Omachi2007","id":"999","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"実環境文字認識のための面積比による付加情報埋込","en":"Supplementary Information Embedment with Area Ratio for Camera-Based Character Recognition"},"journal":{"ja":"電子情報通信学会論文誌D","en":"Trans. IEICE"},"volume":"J90-D","number":"12","pages":"3246-3256","year":2007,"month":12,"review":"reviewed","fields":["pattern-recognition"],"url":"http://search.ieice.org/bin/summary.php?id=j90-d_12_3246","note":{"ja":"論文のPDFファイルを<a href=\"http://www.ieice.org/jpn/trans_online/\">電子情報通信学会 Transactions online</a>からダウンロードすることができます．","en":"PDF file is available at <a href=\"http://www.ieice.org/jpn/trans_online/\">IEICE Transactions online (in Japanese)</a>."},"abstract":{"ja":"ディジタルカメラを入力デバイスとして実環境中の文字を高精度に認識するために，\n文字画像と同時に認識補助のための付加情報を提示する方法が検討されている．\n付加情報は，人間にとって自然な形で提示されること，および，\n幾何学的変形に対してロバストに抽出できることが要求される．\n本論文では，これらの要求を満たす手法として，面積比を利用した付加情報\n提示手法を提案する．\nすなわち，文字パターンを2色で印字することを前提とし，それぞれの色の領域の\n面積比を特定の値とするようにデザインする．\n具体的には，文字に影を付加したり輪郭線を別の色とする．\nこれらは文字パターンのデザインとしてすでに行われており，提案手法は\nその線幅や面積を変えるにすぎない．\n従って，提案手法は様々な用途に広く応用することが可能である．\n面積比はアフィン変換に不変であり，アフィン変換を受けた環境においても\n誤りなく抽出されることが期待される．\n実際に付加情報を埋め込んだ文字パターンを作成し，ディジタルカメラで撮影\nされた画像中の文字パターンから付加情報を抽出する実験\nを行い，提案手法の有効性を確認する．\nまた，付加情報を用いて文字を認識する実験を行い，認識精度が向上することを\n確認する．","en":"In order to achieve a highly accurate recognition of characters in a scene \nimage with a digital camera, there are some attempts on offering\nsupplementary information for recognition with a character image.\nThe information should be human-friendly and robust against geometric\ndistortions.\nIn this paper, we propose a method of embedding information in a character\npattern by designing a character pattern in two colors\nso that the area ratio of the two colors gives the information.\nIt is expected that the area ratio is correctly extracted even if the\ncharacter pattern is affine-transformed since the area ratio is\naffine invariant.\nThe effectiveness of the proposed method is confirmed by\ngenerating character patterns with the embedded information and \ncarrying on an experiment of extracting information and\nrecognizing characters in a scene image taken by a digital camera."}},{"key":"Utsumi2007_Jb","id":"1277","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"内海ゆづ子","en":""},{"ja":"岩井儀雄","en":""},{"ja":"谷内田正彦","en":""}],"title":{"ja":"顔認識のためのウェーブレット特徴量の評価"},"journal":{"ja":"日本知能情報ファジィ学会誌"},"volume":"19","number":"5","pages":"476-487","year":2007,"month":10,"review":"reviewed","fields":[]},{"key":"Uchida2007_IEICE","id":"970","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"酒井 恵","en":"Megumi Sakai"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"オートマトン制御付き最適セグメンテーション法とその実環境文字認識への応用","en":"FSA-Guided Optimal Segmentation and Its Application to Camera-Based Character Recognition"},"journal":{"ja":"電子情報通信学会論文誌D","en":"Trans. IEICE"},"volume":"J90-D","number":"8","pages":"1966-1976","year":2007,"month":8,"review":"reviewed","fields":["pattern-recognition"],"url":"http://search.ieice.org/bin/summary.php?id=j90-d_8_1966","note":{"ja":"論文のPDFファイルを<a href=\"http://www.ieice.org/jpn/trans_online/\">電子情報通信学会 Transactions online</a>からダウンロードすることができます．","en":"PDF file is available at <a href=\"http://www.ieice.org/jpn/trans_online/\">IEICE Transactions online (in Japanese)</a>."},"abstract":{"ja":"本論文では，動的計画法(DP)と有限状態オートマトン(FSA)の組み合わせに基づいた，\n1次元信号の最適セグメンテーション手法を提案する．具体的には，信号の\n性質（例えば信号の値が高い区間と低い区間が交互に繰り返すと言った性質）\nをFSA表現した上で制約条件としてセグメンテーション問題に組み込み，その\n制約下での大\n 局的最適セグメンテーションをDPにより効率的に求める．FSAの導入により，\n信号の性質と一致しないセグメンテーション結果は排除され，精度の向上が見込\nめる．さらに，FSA状態と各区間の対応結果によって各区間の意味づけも可能となる．\n本論文では\n本手法の詳細を述べると共に，さらにある種の実環境文字画像認識タスクに\n適用することでその有効性を評価する．","en":"This paper describes a novel segmentation algorithm for one-dimensional \nsignals. The proposed segmentation algorithm employs an optimization strategy\n based on dynamic programming (DP). In addition, a finite state automaton (FSA) is \nintroduced for incorporating a priori knowledge of the target signal into \nthe optimal segmentation. Specifically, a state of the FSA corresponds\nto a segment having a certain property. For example, we can use \nan FSA with three states, each of which corresponds to a segment whose \naverage signal level is high or middle or low. The relations between \nadjacent segments (such as the relation that a high-level segment is not \nadjacent to a low-level segment but to a middle-level segment) are regulated \nby state transitions. The proposed segmentation algorithm has been applied to \na camera-based character recognition task, where each character pattern is\nprinted with stripes whose widths represent the category of the character.\nThe proposed algorithm could successfully detect the boundary of the stripes\nand provide a recognition rate over 99%."}},{"key":"Iwamura2007_IEICE","id":"958","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"付加情報を用いるパターン認識","en":"Pattern Recognition with Supplementary Information"},"journal":{"ja":"電子情報通信学会論文誌D","en":"Trans. IEICE"},"volume":"J90-D","number":"2","pages":"460-470","year":2007,"month":2,"review":"reviewed","fields":["pattern-recognition"],"url":"http://search.ieice.org/bin/summary.php?id=j90-d_2_460","note":{"ja":"論文のPDFファイルを<a href=\"http://www.ieice.org/jpn/trans_online/\">電子情報通信学会 Transactions online</a>からダウンロードすることができます．","en":"PDF file is available at <a href=\"http://www.ieice.org/jpn/trans_online/\">IEICE Transactions online (in Japanese)</a>."},"abstract":{"ja":"本論文では  パターンが属するクラスの情報(付加情報)をパターンと同時に\n識別器に入力し，パターンと付加情報から矛盾のない答えを導くことで誤認\n識を防ぐ方式を検討する．\nこの方式では付加情報の情報量が増えれば増えるほど認識率は100%に近付く．\nそのため，従来のパターン認識のように，いかに認識性能を向上させるかで\nはなく，ある認識率を達成するために必要な付加情報の情報量をいか\nに小さくできるかが課題となる．\n本論文では付加情報の割り当て方 と認識性能の関係を導き，\n実験によりデモンストレーションする．","en":"In this paper, we introduce a new model in which the classifier\nreceives not only a pattern itself but also supplementary information\nthat assists recognition.\nThis model can greatly improve recognition performance with a few bits\nof information.\nFor practical use of pattern recognition, rejection is one of the most\nimportant operations.\nThe proposed model also can reduce error rates by rejection.\nHow to assign such information is described, and\npossible recognition rates are estimated.\nThe proposed model will enable us to achieve\nan arbitrary recognition rate with an arbitrary rejection rate\nwith certain bits of supplementary information required."}},{"key":"Negishi2007","id":"957","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"根岸 和也","en":"Kazuya Negishi"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"阿曽 弘具","en":"Hirotomo Aso"}],"title":{"ja":"孤立文字認識のための変形を許容する高速なテンプレートマッチング","en":"Fast Deformable Template Matching for Isolated Character Recognition"},"journal":{"ja":"電子情報通信学会論文誌D","en":"Trans. IEICE"},"volume":"J90-D","number":"2","pages":"503-513","year":2007,"month":2,"review":"reviewed","fields":["pattern-recognition"],"url":"http://search.ieice.org/bin/summary.php?id=j90-d_2_503","note":{"ja":"論文のPDFファイルを<a href=\"http://www.ieice.org/jpn/trans_online/\">電子情報通信学会 Transactions online</a>からダウンロードすることができます．","en":"PDF file is available at <a href=\"http://www.ieice.org/jpn/trans_online/\">IEICE Transactions online (in Japanese)</a>."},"abstract":{"ja":"情景中の文字を認識するためには，一般に前処理として文字切り出しが\n必要であるが，既存の手法では文字列を成さない「孤立文字」や隣接し\nた2文字が接触した接触文字などを切り出せないという問題がある．\nまた，情景画像中の文字画像は射影変換の影響で文字画像が歪むことが知られ\nている．\nそこで本論文では，射影変換や字形の違いに対する頑健さを持ち，孤立文字や\n接触文字など，既存の手法では切り出せない文字を切り出すための\n文字切り出し手法を提案する．\n提案手法は変形を許容するテンプレートマッチング法の一種であるが，\n一定の条件の下では変形を許容しない単純なテンプレートマッチング手法\nの計算量を下回るという優れた特徴を持つ．\nまた，テンプレートマッチングであるが故に文字切り出しのみでなく，\n文字認識の大分類も同時に行うことができる．\n提案手法を実際の画像に適用したところ，\n射影変換や字形の違いに対して頑健であり，孤立文字である時計の文字盤等や，\n数式中の接触文字を良好に切り出せることが確認できた．","en":"Though accuracy of character recognition for segmented character\nimage is enough for practical use,\nconventional segmentation methods can be applied to only\nwell-defined problems.\nTherefore, they cannot extract an isolated\ncharacter which does not constitute a character string\nand a touching character which is connected to other characters.\nIn this paper, we propose a fast deformable template matching\nmethod for character recognition.\nThis method enables us to extract and recognize such troublesome\ncharacters simultaneously in smaller computational time than\nnon-deformable template matching method.\nIn addition, the proposed method requires less computational time\nto increase of a template."}},{"key":"Nakai2007_IEICE","id":"971","type":"letter","language":"JA","show":"show","authors":[{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"Webカメラを用いたリアルタイム文書画像検索","en":"Real Time Document Image Retrieval using a Web Camera"},"journal":{"ja":"電子情報通信学会論文誌D","en":"Trans. IEICE"},"volume":"J90-D","number":"8","pages":"2262-2265","year":2007,"month":8,"review":"reviewed","fields":["document-image-retrieval"],"url":"http://search.ieice.org/bin/summary.php?id=j90-d_8_2262","note":{"ja":"論文のPDFファイルを<a href=\"http://www.ieice.org/jpn/trans_online/\">電子情報通信学会 Transactions online</a>からダウンロードすることができます．","en":"PDF file is available at <a href=\"http://www.ieice.org/jpn/trans_online/\">IEICE Transactions online (in Japanese)</a>."}},{"key":"Iwamura2007_BMVC","id":"978","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Tomohiro Nakai","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Improvement of Retrieval Speed and Required Amount of Memory for Geometric Hashing by Combining Local Invariants"},"journal":{"ja":"Proc. 18th British Machine Vision Conference (BMVC2007)"},"volume":"2","pages":"1010-1019","year":2007,"month":9,"location":"Warwick, UK","review":"reviewed","fields":["document-image-analysis","document-image-retrieval"],"abstract":{"ja":"The geometric hashing (GH) is a well-known model-based object\nrecognition technique with good properties both\nin retrieval speed and required amount of memory.\nHowever, it has a significant weak point;\nas the number of objects increases,\nboth retrieval speed and required amount of memory increase in\nthe cubic, fourth or higher order.\nRecently, a new technique ``locally likely arrangement hashing (LLAH)'' \nwhose computational cost is a linear order\nhas been proposed.\nThe objective of the current paper is to reveal how LLAH improves\nthe performance.\nBy comparing GH and LLAH,\nwe describe four primary factors of the performance improvement."}},{"key":"Kise2007_BMVC","id":"979","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Kazuto Noguchi","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Simple Representation and Approximate Search of Feature Vectors for Large-Scale Object Recognition"},"journal":{"ja":"Proc. 18th British Machine Vision Conference (BMVC2007)"},"volume":"1","pages":"182-191","year":2007,"month":9,"location":"Warwick, UK","review":"reviewed","fields":["information-retrieval","object-recognition"],"abstract":{"ja":"This paper presents two methods of large-scale recognition of planar objects\nwith a simple representation and approximate search of local feature vectors.\nA central problem of the use of local feature vectors is the burden of computation\nand memory for finding nearest neighbors. To solve this problem,\nthe proposed methods embody the following: (1) a simple bit representation\nof feature vectors and hashing enable us to fast access with less memory, (2)\napproximate search with query perturbation allows us to find approximate\nnearest neighbors efficiently. From large-scale experiments using 10,000 objects\nin the database and 2,000 query images, it was found that only 10?20%\nof correct nearest neighbors were enough for achieving recognition rate of\n98.0%. The processing time for achieving this rate was 8.3 ms / query (excluding\ntime for feature extraction). We have also tested the scalability of\na proposed method using the database of 100,000 objects and obtained the\nresult of 92.3% accuracy in 4.5 ms /query."}},{"key":"Nakai2007_ICDAR","id":"996","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Tomohiro Nakai","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"A Method of Annotation Extraction from Paper Documents Using Alignment Based on Local Arrangements of Feature Points"},"journal":{"ja":"Proc. 9th International Conference on Document Analysis and Recognition (ICDAR2007)"},"volume":"1","pages":"23-27","year":2007,"month":9,"location":"Curitiba, Brazil","review":"reviewed","fields":["document-image-analysis","document-image-retrieval"],"doi":"10.1109/ICDAR.2007.4378669","abstract":{"ja":"Annotations on paper documents include important information. We can exploit the information by extracting and analyzing annotations. In this paper, we propose a method of annotation extraction from paper documents. Unlike previous methods which limit colors or types of annotations to be extracted, the proposed method attempts to extract annotations by comparing a document image of an annotated document with its original document image for removing the limitations. The proposed method is characterized by fast matching and flexible subtraction of images both of which are essential to the annotation extraction by comparison. Experimental results have shown that color annotations can be extracted from color documents."}},{"key":"Uchida2007_ICDAR","id":"982","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Seiichi Uchida","en":""},{"ja":"Megumi Sakai","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Extraction of Embedded Class Information from Universal Character Pattern"},"journal":{"ja":"Proc. 9th International Conference on Document Analysis and Recognition (ICDAR2007)"},"volume":"1","pages":"437-441","year":2007,"month":9,"location":"Curitiba, Brazil","review":"reviewed","fields":["document-image-analysis"],"doi":"10.1109/ICDAR.2007.4378747","awards":{"en":["IAPR/ICDAR2007 Best Paper Award Winner"]},"abstract":{"ja":"This paper is concerned with a universal pattern, which\nis defined as a character pattern designed to have high\nmachine-readability. This universal pattern is a character\npattern printed with stripes. The cross ratio calculated\nfrom the widths of the stripes represents the character class.\nThus, if the boundaries of the stripes can be detected for\nmeasuring the widths, the class can be determined without\nordinary recognition process. Furthermore, since the cross\nratio is invariant to projective distortions, the correct class\nwill be still determined under those distortions. This paper\ndescribes a practical scheme to recognize this universal\npattern. The proposed scheme includes a novel algorithm to\ndetect the stripe boundaries stably even from the universal\npattern image contaminated by non-uniform lighting and\nnoise. The algorithm is realized by a combination of a dynamic\nprogramming-based optimal boundary detection and\na finite state automaton which represents the property of\nthe universal pattern. Experimental results showed the proposed\nscheme could recognize 99.6% of the universal pattern\nimages which underwent heavy projective distortions\nand non-uniform lighting."}},{"key":"Iwamura2007_CBDAR","id":"993","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Ryo Niwa","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Shinichiro Omachi","en":""}],"title":{"ja":"Rectifying Perspective Distortion into Affine Distortion Using Variants and Invariants"},"journal":{"ja":"Proc. Second International Workshop on Camera-Based Document Analysis and Recognition (CBDAR2007)"},"pages":"138-145","year":2007,"month":9,"location":"Curitiba, Brazil","review":"reviewed","fields":["document-image-analysis"],"abstract":{"ja":"For user convenience, document image processing captured\nwith a digital camera instead of a scanner has been\nresearched. However, existing methods of document image\nprocessing are not usable for a perspective document image\ncaptured by a digital camera because most of them are\ndesigned for the one captured by a scanner. Thus, we have\nto rectify the perspective of the document image and obtain\nthe frontal image as if it was captured by a scanner. In this\npaper, for eliminating perspective distortion from a planar\npaper without any prior knowledge, we propose a new rectification\nmethod of a document image introducing variants\nwhich change according to the gradient of the paper and invariants\nwhich do not change against it. Since the proposed\nmethod does not use strong assumptions, it is widely applicable\nto many document images unlike other methods. We\nconfirmed the proposed method rectifies a document image\nsuffering from perspective distortion and acquires the one\nwith affine distortion."}},{"key":"Nakai2007_CBDARa","id":"994","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Tomohiro Nakai","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Camera Based Document Image Retrieval with More Time and Memory Efficient LLAH"},"journal":{"ja":"Proc. Second International Workshop on Camera-Based Document Analysis and Recognition (CBDAR2007)"},"pages":"21-28","year":2007,"month":9,"location":"Curitiba, Brazil","review":"reviewed","fields":["document-image-analysis","document-image-retrieval"],"abstract":{"ja":"In this paper, we propose improvements of our camera-based document image retrieval method with Locally Likely Arrangement Hashing (LLAH). While LLAH has high accuracy, efficiency and robustness, it requires a large amount of memory. It is also required to speed up the retrieval of LLAH for applications to real-time document image retrieval. For these purposes, we introduce the following two improvements. The first one is reduction of the required amount of memory by removing less important features for indexing from the database and simplifying structure of the database. The second improvement is to reduce exploring alternatives during the retrieval process. From the experimental results, we have confirmed that the proposed improvements realize reduction of the required amount of memory by about 80% and that of processing time by about 60%."}},{"key":"Nakai2007_CBDARb","id":"995","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Tomohiro Nakai","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Real-Time Document Image Retrieval with More Time and Memory Efficient LLAH"},"journal":{"ja":"Proc. Second International Workshop on Camera-Based Document Analysis and Recognition (CBDAR2007)"},"pages":"168-169","year":2007,"month":9,"location":"Curitiba, Brazil","review":"reviewed","fields":["document-image-analysis","document-image-retrieval"],"abstract":{"ja":"We propose a real-time document image retrieval system using a web camera. This is an implementation of the camera based document image retrieval method presented in CBDAR07 oral session as a real-time document image retrieval system."}},{"key":"Furuya2007","id":"1003","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"古谷 嘉男","en":"Yoshio Furuya"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"内田 誠一","en":"Seiichi Uchida"}],"title":{"ja":"付加情報の誤りと誤認識率の関係","en":"Relationship Between Errors of Supplementary Information andMisrecognition Rates"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"presen_id":"PRMU2007-152","pages":"95-100","year":2007,"month":12,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"通常のパターン認識とは異なる，付加情報を用いるパターン認識という新しい枠組みが提案されている．\nこれは，通常のパターンと同時にそれを補助する情報(付加情報) をパターンとは別に用意し，それらを同時に用いる\nことにより誤認識率を減少させる方式である．これまでに，付加情報の復元を誤らないという仮定の下で，誤認識率\nを理論的に最小にする方法が提案されている．しかし，現実世界で付加情報の復元を行うと，観測誤差があるため付\n加情報の復元を誤らないということは稀である．そこで，本稿では，付加情報の復元を誤る際に，どのようにして付\n加情報を与えれば誤認識率を減少させることが可能であるかについて検討を行い，それに基づく実験を行うことで，\n誤認識率が減少する可能性があることを示す．","en":"Pattern recognition with supplementary information is a new pattern recognition framework that determines\nan output class by combining a classifier’s output and supplementary information suggesting the true class.\nUnder the condition that supplementary information does not contain error, we have proposed a theory that reduces\nmisrecognition rates. However, in the real world, we cannot observe any measure without error. Thus, in this\npaper, we discuss how to reduce misrecognition rates using the erroneous supplementary information, and show the\npossibility to reduce misrecognition rates experimentally."}},{"key":"Iwamura2007_IEICEPRMU_Robust","id":"1001","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"阿曽 弘具","en":"Hirotomo Aso"}],"title":{"ja":"次元の縮退を利用した頑健な共分散行列の逆行列の推定法","en":"Robust Estimation of Inverse Covariance Matrix by Shrinkage Technique"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"presen_id":"PRMU2007-156","year":2007,"month":12,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"二次識別関数(QDF)やマハラノビス距離(MD)は，\nパターン認識においてよく用いられる基本的な識別関数(識別器)で，\n各クラスとテストサンプルの類似度(距離)を算出するために用いられる．\nこれらの計算にはパターンの分布を表す共分散行列の逆行列が必要であり，\nその推定精度は認識性能に直結する．\n本研究の目的は，共分散行列の逆行列を高精度で\n推定し，高い認識性能を得ることである．\n我々は既に「次元の縮退」を利用した共分散行列の逆行列の頑健な推定方法を\n提案しており，従来手法よりも頑健な推定が可能であることを実験により確認\nしている．\n本稿では，この手法に球状検定とブロック対角化を導入することにより，\n理論の簡明化および性能向上を実現する．","en":"In pattern recognition,\nimportant measures including quadratic discriminant function (QDF)\nand Mahalanobis distance (MD) depend on \nthe inverse of a covariance matrix.\nSince estimation accuracy of the matrix roughly determines\npattern recognition performance,\nwe have proposed an estimating method using\n``a shrinkage technique.''\nIn this paper, by introducing sphericity test and block\ndiagonalization to the method, we achieve \nsimplification and performance improvement."}},{"key":"Iwamura2007_IEICEPRMU_Attempt","id":"997","type":"domestic_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu IWAMURA","en":"Masakazu IWAMURA"},{"ja":"Takayuki HONDO","en":"Takayuki HONDO"},{"ja":"Kazuto NOGUCHI","en":"Kazuto NOGUCHI"},{"ja":"Koichi KISE","en":"Koichi KISE"}],"title":{"ja":"An Attempt of CUDA Implementation of PCA-SIFT","en":"An Attempt of CUDA Implementation of PCA-SIFT"},"journal":{"ja":"IEICE Technical Report","en":"IEICE Technical Report"},"volume":"107","number":"281","presen_id":"PRMU2007-117","pages":"149-154","year":2007,"month":10,"review":"not_reviewed","fields":["pattern-recognition","object-recognition"],"note":{"ja":"KJPR2007","en":"KJPR2007"},"abstract":{"ja":"GPGPU (General-Purpose computation on GPUs) is a paradigm to use\nGPUs (graphics processing units) for general computation.\nDue to recent remarkable improvement of GPU, GPU outperforms CPU\nin computation ability.\nHowever, most people could not use the ability for general computation\nbecause existing programming languages require knowledge about GPU\nhardware architectures and computer graphics for GPGPU computing.\nRecently, a new GPU language CUDA (Compute Unified Device\nArchitecture) has been released from NVIDIA.\nThe CUDA code is C language style and \nhas less computational restriction.\nThus, usual operations of C language can run on GPU without much\nspecial knowledge.\nIn this report, we briefly introduce CUDA language programming and\nreport a CUDA implemented of PCA-SIFT.\nCompared to a CPU implementation,\nour CUDA implementation reduced the processing time to around 1/4.\nIn addition, we also report an interesting phenomenon results useful\nfor practical use of CUDA.","en":"GPGPU (General-Purpose computation on GPUs) is a paradigm to use\nGPUs (graphics processing units) for general computation.\nDue to recent remarkable improvement of GPU, GPU outperforms CPU\nin computation ability.\nHowever, most people could not use the ability for general computation\nbecause existing programming languages require knowledge about GPU\nhardware architectures and computer graphics for GPGPU computing.\nRecently, a new GPU language CUDA (Compute Unified Device\nArchitecture) has been released from NVIDIA.\nThe CUDA code is C language style and \nhas less computational restriction.\nThus, usual operations of C language can run on GPU without much\nspecial knowledge.\nIn this report, we briefly introduce CUDA language programming and\nreport a CUDA implemented of PCA-SIFT.\nCompared to a CPU implementation,\nour CUDA implementation reduced the processing time to around 1/4.\nIn addition, we also report an interesting phenomenon results useful\nfor practical use of CUDA."}},{"key":"Kise2007_IEICEPRMU","id":"998","type":"domestic_conf","language":"EN","show":"show","authors":[{"ja":"Koichi KISE","en":"Koichi KISE"},{"ja":"Satoshi YOKOTA","en":"Satoshi YOKOTA"},{"ja":"Akira SHIOZAKI","en":"Akira SHIOZAKI"}],"title":{"ja":"Copyright Protection of Images Based on Large-Scale Image Recognition","en":"Copyright Protection of Images Based on Large-Scale Image Recognition"},"journal":{"ja":"IEICE Technical Report","en":"IEICE Technical Report"},"volume":"107","number":"281","presen_id":"PRMU2007-91","pages":"3-8","year":2007,"month":10,"review":"not_reviewed","fields":["pattern-recognition"],"note":{"ja":"KJPR2007","en":"KJPR2007"},"abstract":{"ja":"This paper presents a method of copyright protection\nfor digital still images based on their recognition\nwith a large-scale image database.\nFor the purpose of copyright protection of images，\ndigital watermarking has been extensively studied for more than a decade.\nHowever，it is still difficult to achieve enough robustness against\nalternations such as StirMark attacks.\nIn order to solve this problem, we propose a method of copyright protection\nbased on new image recognition technologies that are tolerant to various\nimage distortions including geometric transformations.\nThe image recognition enables us to find a corresponding image\nfrom at least tens of thousands of reference images to be protected\nin response to a query image\nthat undergoes StirMark attacks and digital watermarking.\nFrom experimental results using 10,000 images in the database，\nwe have confirmed that the method is capable of recognizing\n99.5\\% of query images \nonly within 216 ms / query (excluding feature extraction).","en":"This paper presents a method of copyright protection\nfor digital still images based on their recognition\nwith a large-scale image database.\nFor the purpose of copyright protection of images，\ndigital watermarking has been extensively studied for more than a decade.\nHowever，it is still difficult to achieve enough robustness against\nalternations such as StirMark attacks.\nIn order to solve this problem, we propose a method of copyright protection\nbased on new image recognition technologies that are tolerant to various\nimage distortions including geometric transformations.\nThe image recognition enables us to find a corresponding image\nfrom at least tens of thousands of reference images to be protected\nin response to a query image\nthat undergoes StirMark attacks and digital watermarking.\nFrom experimental results using 10,000 images in the database，\nwe have confirmed that the method is capable of recognizing\n99.5\\% of query images \nonly within 216 ms / query (excluding feature extraction)."}},{"key":"Horimatsu2007_MC","id":"981","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"堀松 晃","en":"Akira Horimatsu"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"局所特徴の組み合わせに基づく相似不変な文字認識","en":"Similarity-Invariant Character Recognition Based on Combination of Local Features"},"journal":{"ja":"電気学会 電子・情報・システム部門大会講演論文集"},"presen_id":"MC3-4","year":2007,"month":9,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Sezaki2007","id":"990","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"瀬崎 直人","en":"Naoto Sezaki"},{"ja":"竹内 謹治","en":"Kinji Takeuchi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"タグの共起と画像の類似性に基づくタグ付け支援システム","en":"Tagging System Based on Co-occurrence of Tags and Similarity of Images"},"journal":{"ja":"電気学会 電子・情報・システム部門大会講演論文集"},"presen_id":"MC9-3","year":2007,"month":9,"review":"not_reviewed","fields":["information-retrieval"]},{"key":"Tanimoto2007","id":"991","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"谷本 真吾","en":"Shingo Tanimoto"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"PDF文書からの手順記述抽出に関する一検討","en":"Investigation of a System for Extracting Procedures from PDF Documents"},"journal":{"ja":"電気学会 電子・情報・システム部門大会講演論文集"},"presen_id":"MC9-4","year":2007,"month":9,"review":"not_reviewed","fields":["information-retrieval"]},{"key":"Yokota2007","id":"992","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"横田 哲","en":"Satoshi Yokota"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"汐崎 陽","en":"Akira Shiozaki"}],"title":{"ja":"画像認識を用いた著作権保護システムの実験的検討","en":"Experimental Investigation of a Copyright Protection System Based on Image Recognition"},"journal":{"ja":"電気学会 電子・情報・システム部門大会講演論文集"},"presen_id":"MC3-3","year":2007,"month":9,"review":"not_reviewed","fields":["pattern-recognition","document-image-analysis"]},{"key":"Nakai2007_MIRU","id":"988","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"LLAHによるリアルタイム文書画像検索の日本語文書への拡張","en":"Extension of Real-Time Document Image Retrieval with LLAH for Japanese Documents"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2007)論文集","en":"Proceedings of MIRU 2007"},"presen_id":"DS-18","pages":"471-472","year":2007,"month":8,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"],"abstract":{"ja":"著者らは既に，英文文書を対象としたリアルタイム文書画像検索法を提案している．この手法は，LLAHと呼ばれる高速かつ高精度な画像認識手法を用いることで，大規模データベースからの文書画像検索をリアルタイムで実現したものである．しかし，LLAHは特徴点の配置のみを手掛かりとして検索を行うため，その精度が特徴点の識別性・安定性に大きく依存するという問題がある．そのため，この手法の対象は単語の重心という特徴的かつ安定な特徴点が得られる英文文書に限られていた．本稿では，LLAHの特徴量計算方法を改良し，識別性や安定性に問題のある特徴点からでも高精度な検索を可能にした画像認識法を提案する．提案手法により，従来困難であった日本語文書の検索が実現できる．発表では，日本語文書を対象としたリアルタイム文書画像検索のデモを行う．","en":"We have already proposed a real-time retrieval method for English documents. It is realized with a fast and accurate image recognition method called LLAH. Since LLAH exploits only arrangements of feature points, its accuracy depends heavily on discrimination power and stability of feature points extracted from images. Therefore the method has been applicable to English documents which give centroids of word regions as discriminative and stable feature points. In this paper, we propose an improved LLAH which enables to retrieve images with less discriminative and less stable feature points such as those extracted from Japanese documets. In the presentation, we show a demonstration of real-time retrieval of Japanese documents."}},{"key":"Utsumi2007_MIRU","id":"1295","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"内海ゆづ子","en":""},{"ja":"岩井儀雄","en":""},{"ja":"谷内田正彦","en":""}],"title":{"ja":"顔認識のための特徴量の評価"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2007)"},"pages":"576-581 (IS-1-18)","year":2007,"month":7,"review":"not_reviewed","fields":[]},{"key":"Iwamura2007_MIRU","id":"987","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"特徴点の配置に基づく画像検索手法の解析 —Geometric HashingとLLAHの比較—","en":"An Analysis of Image Retrieval Methods Based on Arrangements of Feature Points—Comparison Between Geometric Hashing and LLAH—"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2007)論文集","en":"Proceedings of MIRU 2007"},"presen_id":"IS-4-10","pages":"1200-1205","year":2007,"month":7,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"],"abstract":{"ja":"Geometric Hashingは広く知られた特徴点ベースの物体認識手法である．\n検索したい特徴点の配置をデータベースから全探索よりも高速に検索する\nことができる．\n一方，我々は\n特徴点の局所的配置に基づく高精度，省メモリかつ高速な検索手法である\n``Locally Likely Arrangement Hashing (LLAH)''\nを提案している．\nいずれの手法も特徴点の照合を高速に行うという点で共通しているが，\n検索速度とメモリ使用量の使用量に関してはLLAHが優れている．\n本稿ではLLAHをGeometric Hashingと比較することにより，どのような\n要因がLLAHの精度を保った高速化に寄与しているかを考察する．\nまた，LLAHの検索性能は検索対象の撮影角度に依存するため，\n撮影角度とLLAHの検索性能の関係についても考察する．","en":"The geometric hashing (GH) is a well-known object recognition\ntechnique based on arrangement of feature points.\nRecently, we have proposed\n``locally likely arrangement hashing (LLAH)''\nwhich outperforms GH in both processing time and required amount of\nmemory with little loss of retrieval accuracy.\nOne of two objectives of the current paper is to reveal where the\noutstanding performance of LLAH comes from.\nThe other objective is to discuss the robustness of LLAH\nagainst change of the picture angle which can spoil the performance."}},{"key":"Niwa2007_MIRU","id":"983","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"丹羽 亮","en":"Ryo Niwa"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"}],"title":{"ja":"変量と不変量に基づく射影歪みからアフィン歪みへの補正","en":"Rectifying Perspective Distortion into Affine Distortion Using Variants and Invariants"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2007)論文集","en":"Proceedings of MIRU 2007"},"presen_id":"IS-2-03","pages":"720-725","year":2007,"month":7,"review":"not_reviewed","fields":["document-image-analysis"],"abstract":{"ja":"利用者の利便性向上の観点から，デジタルカメラをスキャナの代わりに用いて文書画像処理を行う試みが\n盛んに研究されている．しかし，既存の文書画像処理技術の多くはスキャナで取得した文書画像を対象としているた\nめ，デジタルカメラで斜めから撮影した文書画像には適用できない．このような場合，文書画像に生じた射影歪みを\n補正し，スキャナで取得したかのような真正面の文書画像を得る必要がある．本稿では，紙面の傾きに応じて値が変\n化する「変量」と変化しない「不変量」を組み合わせた文書画像の傾き補正手法を提案する．提案手法は，従来手法\nのように撮影方法やレイアウトに強い制約を用いないため，これまで補正不可能であった文書画像を復元することが\nできる．実験により，射影歪みを受けた文書画像をアフィン歪み程度に復元できることを確認した．","en":"For user convenience, document image processing (DIP) captured with a digital\ncamera instead of a scanner has been researched.\nHowever, existing methods of DIP are not usable for a perspective document\nimage captured by a digital camera since most of them are designed for\nthe one captured by a scanner.\nThus, we have to rectify the perspective of the document image\nand obtain the frontal image as if it was captured by a scanner.\nIn this paper, we propose a new rectification method of a document image\nintroducing variants which change according to the gradient of\nthe paper and invariants which do not change against it.\nSince the proposed method does not use strong assumptions,\nit is widely applicable to many document images unlike other methods.\nWe confirmed the proposed method rectifies a document image suffering\nfrom perspective distortion and acquires the one with Affine distortion."}},{"key":"Noguchi2007_MIRU_Efficient","id":"985","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"野口 和人","en":"Kazuto NOGUCHI"},{"ja":"黄瀬 浩一","en":"Koichi KISE"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"}],"title":{"ja":"近似最近傍探索の多段階化による物体の高速認識","en":"Efficient Recognition of Objects by Cascading Approximate Nearest Neighbor Searchers"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2007)論文集","en":"Proceedings of MIRU 2007"},"presen_id":"OS-B2-02","pages":"111-118","year":2007,"month":7,"review":"not_reviewed","fields":["pattern-recognition","object-recognition"],"abstract":{"ja":"SIFT などの局所記述子の最近傍探索によって物体認識を行う場合には，一画像あたりの特徴ベクトルの数が膨大になるため，最近傍探索の効率が重要となる．本稿では，「認識に必要な最近傍探索の精度は画像によって異\nなる」という観点から処理を削減した効率的認識法を提案する．具体的には，近似最近傍探索に基づく識別器を多段階に縦列接続することにより，認識に用いる近似の程度を画像に応じて変更し，大幅な効率化を実現する．一万画像のデータベースを用いた実験の結果，処理時間を，多段階化を行わなかった場合の約1/10，ANN やLSH を近似最近傍探索の手法として用いた場合の約1/40 に削減でき，例えば，認識率98%，処理時間1ms/query を達成できることがわかった．また，8.6%のリジェクトを行うことによって，誤認識率を0%に抑えることもできた．さらに，十万画像を用いた実験から，高いスケーラビリティがあることもわかった．","en":"For object recognition based on nearest neighbor search of local descriptors such as SIFT, it is important to keep the nearest neighbor search efficient to deal with a huge number of descriptors. In this report we propose a new method of efficient recognition based on the observation that the level of accuracy of nearest neighbor search for correct recognition depends on images to be recognized. The proposed method is characterized by the mechanism that multiple recognizers with approximate nearest neighbor search are cascaded in the order of the\nlevel of approximation so as to improve the efficiency by adaptively ontrolling the level to be applied depending on images. From experimental results with 10,000 images, we have confirmed that the proposed method is capable of achieving a recognition rate of 98% in 1 ms / query, which is 1/10 of the recognition time without the cascade,\nand 1/40 of the recognition time with conventional approximate nearest neighbor search such as ANN and LSH. In\naddition, a recognition error rate of the proposed method has been suppressed to 0% by allowing a rejection rate of\n8.6%. Experimental results with 100,000 images show high scalability of the proposed method."}},{"key":"Noguchi2007_MIRU_Real","id":"986","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"野口 和人","en":"Kazuto NOGUCHI"},{"ja":"本道 貴行","en":"Takayuki HONDO"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"},{"ja":"黄瀬 浩一","en":"Koichi KISE"}],"title":{"ja":"近似最近傍探索の多段階化によるリアルタイム物体認識システム","en":"Real-Time Recognition of Objectsby Cascading Approximate Nearest Neighbor Searchers"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2007)論文集","en":"Proceedings of MIRU 2007"},"presen_id":"DS-16","pages":"467-468","year":2007,"month":7,"review":"not_reviewed","fields":["pattern-recognition","object-recognition"],"awards":{"ja":["MIRU2007 デモセッション賞受賞"],"en":["MIRU2007 Best Demonstration Award Winner"]},"abstract":{"ja":"本稿では，局所記述子としてPCA-SIFT を用いたリアルタイム物体認識システムを提案する．従来，\nPCA-SIFT を用いて物体を認識する場合，局所記述子の数が膨大となるため，局所記述子の抽出，ならびに局所記述子の照合にかかる処理時間が問題であった．抽出の問題を解決するため，提案システムでは，GPU(Graphical Processing Unit) を用いる．照合の問題に対しては，多段階化された近似最近傍探索器を用いる解決策を提案する．これら２つの要因によって，例えば，十万物体のデータベースを用いた場合でも，リアルタイム認識が可能となる．","en":"We demonstrate a system for real-time object recognition which employs PCA-SIFT as local descriptors of images. Object recognition based on PCA-SIFT has posed problems of processing time for both extraction and matching of local descriptors because of their large number. To solve the problem about extraction, the proposed\nsystem is equipped with a GPU (Graphical Processing Unit). For the problem about matching, we utilize a cascaded approximate nearest neighbor searchers. These two factors enable us real-time object recognition even with\na database of 100,000 objects."}},{"key":"Sakai2007_MIRU","id":"980","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"酒井 恵","en":"Megumi Sakai"},{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"変量と不変量を用いた文書画像の事例ベース傾き補正","en":"Instance-Based Skew Detection of Document Images by a Combination of Variant and Invariant"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2007)論文集","en":"Proceedings of MIRU 2007"},"presen_id":"IS-4-01","pages":"1147-1152","year":2007,"month":7,"review":"not_reviewed","fields":["document-image-analysis","pattern-recognition"],"abstract":{"ja":"ある特定の幾何変形に対する「変量」と「不変量」のみを利用することで変形量を推定する方法を提案す\nる．特定の幾何変形としてスキューすなわち回転を例に取り説明する．まずあらかじめ様々な文字フォントを回転さ\nせながら回転変量（例えば，文字の外接矩形の面積）の変化を測定し，これを回転不変量（例えば，文字凸包内の黒\n画素と白画素の面積）と共に事例として記憶しておく．そして補正したい画像の各連結成分（多くは単文字に相当）\nから算出した回転不変量により事例を呼び出し，同じく算出した回転変量から文字の回転角を得る．推定された回転\n角により投票を行い，最も投票結果の多いものを推定結果とする．本手法には，文字が直線的な部分形状と並びを持\nつという，従来一般に利用されてきた仮定が不要になる．文書画像55 サンプルを用いて，簡単な傾き推定実験を行っ\nたところ，45 サンプルにおいて誤差1 度以下，54 サンプルにおいて誤差2 度以下という精度を得た．","en":"A novel deformation estimation technique is proposed and applied to document skew estimation. The\nproposed method has two properties. First, it utilizes an invariant and a variant of a target deformation to be\nestimated. Second, it is an instance-based method where the deformation is estimated by referring stored instances\nwhich describe the relation among the deformation, the variant, and the invariant. The result of a skew estimation\nexperiment on 55 document images has shown that the skew angles of 54 document images were successfully\nestimated with errors smaller than 2.0 degrees."}},{"key":"Nakai2007_SSII","id":"977","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"点パターンの高速検索技術に基づくリアルタイム文書画像認識","en":"Real-Time Document Image Recognition Based on Fast Point Pattern Retrieval"},"journal":{"ja":"第13回画像センシングシンポジウム(SSII07)予稿集"},"presen_id":"LD2-07","year":2007,"month":6,"review":"not_reviewed","fields":["document-image-analysis","pattern-recognition","document-image-retrieval"],"abstract":{"ja":"物体認識の手法として，認識対象を撮影した画像から特徴点を抽出し，特徴点の配置に基づいて認識するものが提案されている．しかし，従来手法には，画像に含まれる特徴点の数が多いと処理時間が膨大になるという問題がある．我々はすでに，特徴点の局所的配置を利用することで多数の特徴点のパターンとして記述される画像を高速に認識する手法を提案している．また，この手法の高速性を利用した拡張として，英文文書を対象としたリアルタイム文書画像検索法についても提案している．本稿では，この手法の実用性を確認するために，様々な条件でWebカメラを用いたリアルタイム文書画像検索の実験を行った．その結果，認識対象の一部からでもほぼ100%での検索が可能であり，低い解像度や紙面の湾曲に対するロバスト性をもつことが示された．"}},{"key":"Noguchi2007_SSII","id":"984","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"野口 和人","en":"Kazuto Noguchi"},{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"近似最近傍探索を用いた省メモリ高速物体認識法","en":"High Speed and Small Memory Object Recognition by Approximate Nearest Neighbor Search"},"journal":{"ja":"第13回画像センシングシンポジウム(SSII07)予稿集"},"presen_id":"IN2-23","year":2007,"month":6,"review":"not_reviewed","fields":["pattern-recognition","object-recognition"],"abstract":{"ja":"SIFT 等の画像の局所記述子を用いて，最近傍探索によって物体認識を行う場合には，一画像あたりの特徴ベクトルの数が膨大になるため，探索効率が重要となる．本稿では，「認識のためには大半の最近傍は不要で\nある」という観点から，省メモリで高速性をもった認識法を提案する．一万画像のデータベースを用いた実験の結果，処理時間が従来手法であるANN やLSH を近似最近傍探索の手法として用いた場合の約1/10 に削減でき，例えば，認識率98%，処理時間4.5ms/query\nを達成できることが分かった．"}},{"key":"Noguchi2007_PRMU","id":"1002","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"野口 和人","en":"Kazuto NOGUCHI"},{"ja":"黄瀬 浩一","en":"Koichi KISE"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"}],"title":{"ja":"近似最近傍探索を用いた物体認識のための多段階化とその効果","en":"Cascading Approximate Nearest Neighbor Searchers and Its Effects onObject Recognition"},"journal":{"ja":"電子情報通信学会技術研究報告"},"presen_id":"PRMU2007-44","pages":"99-104","year":2007,"month":6,"review":"not_reviewed","fields":["pattern-recognition","object-recognition"],"abstract":{"ja":"SIFT などの局所記述子の最近傍探索によって物体認識を行う場合には，一画像あたりの特徴ベクトルの\n数が膨大になるため，最近傍探索の効率が重要となる．本稿では，「認識に必要な最近傍探索の精度は画像によって異\nなる」という観点から処理を削減した効率的認識法を提案する．具体的には，近似最近傍探索に基づく識別器を多段\n階に縦列接続することにより，認識に用いる近似の程度を画像に応じて変更し，大幅な効率化を実現する．一万画像\nのデータベースを用いた実験の結果，処理時間を，多段階化を行わなかった場合の約1/5，ANN やLSH を近似最近\n傍探索の手法として用いた場合の約1/40 に削減でき，例えば，認識率98%，処理時間1ms/query を達成できること\nが分かった．また，12.2%のリジェクトを行うことによって，誤認識率を0.25%に抑えることもできた．","en":"For object recognition based on nearest neighbor search of local descriptors such as SIFT, it is important\nto make the nearest neighbor search efficient to deal with a huge number of descriptors. In this report we\npropose a new method of efficient recognition based on the observation that the level of accuracy of nearest neighbor\nsearch for correct recognition depends on images to be recognized. The proposed method is characterized by the\nmechanism that multiple recognizers with approximate nearest neighbor search are cascaded in the order of the\nlevel of approximation so as to improve the efficiency by adaptively controlling the level to be applied depending on\nimages. From experimental results for a database of 10,000 images, we have confirmed that the proposed method is\ncapable of achieving a recognition rate of 98% in 1 ms / query, which is about 1/5 of the recognition time without\nthe cascade, and 1/40 of the recognition time with conventional methods of approximate nearest neighbor search\nsuch as ANN and LSH. In addition, a recognition error rate of the proposed method has been suppressed to 0.25%\nby allowing a rejection rate of 12.2%."}},{"key":"Nakai2007_PRMU","id":"965","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"特徴点の局所的配置に基づく位置合わせを用いた文書からの書き込み抽出法","en":"A Method of Annotation Extraction from Documents Using Alignment Based on Local Arrangements of Feature Points"},"journal":{"ja":"電子情報通信学会技術研究報告"},"volume":"106","presen_id":"PRMU2006-245","pages":"61-66","year":2007,"month":3,"review":"not_reviewed","fields":["document-image-analysis","pattern-recognition","document-image-retrieval"],"abstract":{"ja":"紙媒体の文書における書き込みには重要な情報が含まれているため，書き込みを抽出し，解析することでその情報を活用することが望まれている．本稿では，紙文書からの書き込み抽出の手法を提案する．これは，スキャナによって得られた書き込みの含まれる文書画像と元となった文書画像を比較することで書き込みを取り出すものである．提案手法の特徴は，高精度な位置合わせと柔軟な差分処理である．特徴点の局所的配置に基づく検索法によって得られる特徴点の対応関係に基づいて位置合わせを行い，撮像条件の変化によって生じる歪みに対してロバストな差分処理によってノイズの少ない書き込み画像を取得する．実験により，提案手法によってカラー文書の76%，モノクロ文書の94%から書き込みの抽出が可能であることが示された．","en":"Annotations on paper documents include important information, which can be exploited by their extraction and analysis. In this report, we propose a method of annotation extraction from paper documents. In the proposed method, annotations are extracted by comparing a document image of an annotated document captured using a scanner and its original document image. The proposed method is characterized by precise matching and flexible subtraction of images. Experimental results have shown that annotations can be extracted using the proposed method with 76% accuracy in the cases of color documents and 94% accuracy in the cases of monochrome documents."}},{"key":"Sakai2007_IEICEPRMU","id":"973","type":"domestic_conf","language":"JA","show":"show_except_abstract","authors":[{"ja":"酒井 恵","en":"Megumi Sakai"},{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"変量と不変量の組合わせによる文書画像の傾き補正","en":"Skew Detection of Document Images by a Combination of Variant and Invariant"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"106","presen_id":"PRMU2006-251","pages":"97-102","year":2007,"month":3,"review":"not_reviewed","fields":["document-image-analysis","pattern-recognition"]},{"key":"Tanaka2007","id":"972","type":"domestic_conf","language":"JA","show":"show_except_abstract","authors":[{"ja":"田中 一弘","en":"Kazuhiro Tanaka"},{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"データ埋め込みペン実現に向けたデータ抽出精度の検証","en":"Evaluation of Data Extraction Accuracy toward the Realization of Data-embedding Pen"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"106","presen_id":"PRMU2006-247","pages":"73-78","year":2007,"month":3,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Takeuchi2007_SIG","id":"962","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"竹内 謹治","en":"Kinji Takeuchi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"Earth Mover's Distanceに基づくText-Based Image Retrieval","en":"Text-Based Image Retrieval Using Earth Mover's Distance"},"journal":{"ja":"情報処理学会研究報告"},"number":"NL-177-5","year":2007,"month":1,"review":"not_reviewed","fields":["information-retrieval"],"abstract":{"ja":"本稿ではEarth Mover's Distance(EMD)を用いてキーワードによる画像検索(TBIR)を行う手法を提案する．\n従来のTBIR手法の性能を向上させるアプローチの一つとして，キーワードに加えて画像自体から得られる特徴量を利用することが考えられる．\nEMDを用いることで，本来，異質であるキーワードと画像特徴量を統一的に扱うことが可能である．\n提案手法の有効性を確認するため，12295枚の画像を対象にして，LSI，PLSIなどの従来手法との比較実験を行った結果，\n平均適合率において，最良の結果が得られた．また，EMDを用いる手法の高速化のために，近似方法を検討した結果，LSIと同程度の検索性能を保ちながら，理想的には計算量を約8%に削減できることが分かった．","en":"In this report, we propose a method of Text-Based Image Retrieval(TBIR)  using Earth Mover's Distance(EMD). \nOur approach uses EMD to combine  textual and visual features into one semantic space,\nto improve the performance of TBIR.\nWe compare our method with four conventional methods including LSI and PLSI.\nThe results show our method is superior to others in average precision.\nWe also applied an approximation in computing EMD for improving the efficiency of the proposed method.\nThe results show that the approximate method can reduce the computational complexity in about 8%,\nwith little loss of average precision."}},{"key":"Takeuchi2007_MThesis","id":"975","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"竹内 謹治","en":""}],"title":{"ja":"Earth Mover's Distanceを利用したキーワードによる画像検索"},"year":2007,"month":3,"review":"not_reviewed","fields":["information-retrieval"]},{"key":"Horimatsu2007_BThesis","id":"966","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"堀松 晃","en":"Akira Horimatsu"}],"title":{"ja":"局所的な形状情報に基づく相似不変な文字認識"},"year":2007,"month":3,"review":"not_reviewed","fields":[]},{"key":"Miki2007","id":"1035","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"三木 博史","en":""}],"title":{"ja":"DBNを用いた人物行動に基づく物体認識"},"year":2007,"month":3,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Miyake2007","id":"976","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"三宅弘志","en":""}],"title":{"ja":"局所特徴量を用いた動画像とランドマークの対応付け"},"year":2007,"month":3,"review":"not_reviewed","fields":[]},{"key":"Niwa2007_BThesis","id":"967","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"丹羽 亮","en":""}],"title":{"ja":"変量と不変量に基づく紙面の射影歪み補正"},"year":2007,"month":3,"review":"not_reviewed","fields":[]},{"key":"Noguchi2007_BThesis","id":"968","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"野口 和人","en":"Kazuto Noguchi"}],"title":{"ja":"最近傍探索の近似と多段階化による物体の高速認識"},"year":2007,"month":3,"review":"not_reviewed","fields":["object-recognition"]},{"key":"Nouso2007","id":"969","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"能祖 正樹","en":"Masaki Nouso"}],"title":{"ja":"PDF文書からの手順とその属性の抽出"},"year":2007,"month":3,"review":"not_reviewed","fields":[]},{"key":"Tominaga2007","id":"974","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"富永 昌文","en":""}],"title":{"ja":"部分領域の色情報と局所特徴量を用いた画像の索引付け"},"year":2007,"month":3,"review":"not_reviewed","fields":[]},{"key":"Nakai2007","id":"1955","type":"survey","language":"JA","show":"show","authors":[{"ja":"中居 友弘","en":""},{"ja":"黄瀬 浩一","en":""},{"ja":"岩村 雅一","en":""}],"title":{"ja":"おめでとう論文賞：特徴点の局所的配置に基づくデジタルカメラを用いた高速文書画像検索"},"journal":{"ja":"電子情報通信学会 情報・システムソサイエティ誌"},"volume":"12","number":"3","pages":"14","year":2007,"month":11,"review":"not_reviewed","fields":[]},{"key":"Kise2007","id":"1016","type":"survey","language":"JA","show":"show_except_datafile","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"大阪府立大学　大学院工学研究科　知能情報工学分野　知能メディア処理研究室"},"journal":{"ja":"システム/制御/情報"},"volume":"51","number":"7","pages":"313-314","year":2007,"month":7,"review":"not_reviewed","fields":[]},{"key":"Kise2007_IL","id":"963","type":"survey","language":"JA","show":"show_except_datafile","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"情報埋め込みによる高精度文字パターン認識 —ユニバーサル・パターンの実現に向けて—"},"journal":{"ja":"画像ラボ","en":"Image Lab"},"volume":"18","number":"4","pages":"18-22","year":2007,"month":4,"publisher":"日本工業出版","review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"カメラで撮影した文字や、自由に手書きをした文字を、人間と同程度の精度を保ちつつ手軽に入力することは、機械による認識にとって未だに実現されていない大きな課題の一つである。本稿では、認識対象の文字パターンに、付加情報と呼ぶ認識を助ける情報を埋め込んで利用することにより、従来の文字認識と比べてより高精度な認識を実現する方法について述べる。我々は、このような文字パターンを、人間にも機械にも読めるという意味でユニバーサル・パターンと名付けている。ユニバーサル・パターンを用いると人間のための文字と機械のためのバーコードを併記する必要がなくなるため、外観の改善など種々の利点を得ることができる。"}},{"key":"Matsushita2007","id":"1623","type":"other","language":"JA","show":"show_except_datafile","authors":[{"ja":"松下光範","en":""},{"ja":"山西良典","en":""},{"ja":"松井勇佑","en":""},{"ja":"岩田基","en":""},{"ja":"上野未貴","en":""},{"ja":"西原陽子","en":""},{"ja":"中村聡史","en":""}],"title":{"ja":"私のブックマーク「コミック工学（Comic Computing）」"},"journal":{"ja":"人工知能学会誌"},"volume":"32","number":"6","pages":"999-1007","year":2007,"month":11,"fields":[]},{"key":"Utsumi2007_Ja","id":"1282","type":"other","language":"JA","show":"show","authors":[{"ja":"内海ゆづ子","en":""},{"ja":"岡崎直観","en":""}],"title":{"ja":"学生フォーラム「実世界で役立つ言語処理へ」黒橋禎夫氏"},"journal":{"ja":"人工知能学会誌"},"volume":"22","number":"4","pages":"577-579","year":2007,"month":7,"review":"not_reviewed","fields":[]},{"key":"Nakai2006_IEICE","id":"847","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"特徴点の局所的配置に基づくデジタルカメラを用いた高速文書画像検索","en":"Fast Camera-Based Document Image Retrieval Using Local Arrangements of Feature Points"},"journal":{"ja":"電子情報通信学会論文誌D","en":"Trans. IEICE"},"volume":"J89-D","number":"9","pages":"2045-2054","year":2006,"month":9,"review":"reviewed","fields":["document-image-retrieval"],"url":"http://search.ieice.org/bin/summary.php?id=j89-d_9_2045","awards":{"ja":["第63回（平成18年度） 電子情報通信学会 論文賞受賞"],"en":["2006 IEICE Best Paper Award Winner"]},"abstract":{"ja":"本論文では，デジタルカメラで撮影して得られた\n文書画像をもとに，データベースから対応する文書画像を見つける\n文書画像検索の手法を提案する．\n従来のスキャナを利用した文書画像検索と異なり，デジタルカメラを利用する\n場合には射影歪みなどのデジタルカメラ特有の問題に対処する必要がある．\nまた，大規模データベースでの利用を可能にするため，\n検索の効率性も重要である．\nこれらの問題に対処するため，本論文では\n特徴点の局所的配置を特徴量とし，\nハッシュ表を用いた投票によって検索する手法を提案する．\n本手法では，射影歪みを実用的な範囲に制限し，\n特徴点の組み合わせを局所領域に限定することで，計算量の削減を実現する．\nまた，高精度な検索を実現するため，射影変換の不変量である複比で\n特徴点の配置を表現し，特徴量としている．\n実験により，10,000ページのデータベースから精度98%，\n平均処理時間約160msec.（特徴点抽出の画像処理に要する約1sec.を除く）\nで検索できることが示された．","en":"Document image retrieval is a task of searching document images\nfrom the database based on users' queries.\nIn this paper, we propose a method of document image retrieval which\nemploys document images captured by digital cameras as queries.\nUnlike document image retrieval using scanners,\nit is necessary to deal with the problem of perspective distortion.\nIn the proposed method, document images are characterized using\nlocal arrangements of feature points, which allows us fast retrieval\nof O(N) where N is the number of feature points.\nFor the description of arrangements, perspective invariants\nare employed for the retrieval under perspective distortion."}},{"key":"Nakai2006_IEEJ","id":"952","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"近藤 伸幸","en":"Nobuyuki Kondo"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"松本 啓之亮","en":"Keinosuke Matsumoto"}],"title":{"ja":"情報リサイクルのための文書の書き込み解析","en":"Analysis of Annotation on Documents for Recycling Information"},"journal":{"ja":"電気学会論文誌（Ｃ）","en":"IEEJ Transactions on Electronics, Information and Systems"},"volume":"126","number":"4","pages":"443-450","year":2006,"month":4,"review":"reviewed","fields":[]},{"key":"Uchida2006_IEICE","id":"375","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"カメラによる文字認識のためのカテゴリー情報の埋込に関する検討","en":"Category Data Embedding for Camera-Based Character Recognition"},"journal":{"ja":"電子情報通信学会論文誌D","en":"Trans. IEICE"},"volume":"J89-D","number":"2","pages":"344-352","year":2006,"month":2,"review":"reviewed","fields":["pattern-recognition","document-image-analysis"],"url":"http://search.ieice.org/bin/summary.php?id=j89-d_2_344","abstract":{"ja":"本研究は，バーコードと同程度の精度で? 次元実環境中の文字パターンを認識することを目標とし\nている．実環境中の文字パターンは，撮影状況により様々な歪み，例えば射影変換歪みを受ける．このため，通\n常の文字認識手法の延長線上でこの目標を達成しようとしても，相当の困難が予想される．そこで本論文では，\n文字そのものに機械可読性を補強するような情報を埋め込む方式を検討する．具体的には，文字画像に対し，縞\n模様状のパターンを埋め込む．このパターンを構成する各縞の幅から計算される複比は，文字パターンがどのよ\nうに射影変換歪みを受けてたとしても常に一定値となる．従って，カテゴリと複比の値を予め対応付けておけば，\n抽出された複比を識別の手がかりとして認識時に利用できる．シミュレーション実験の結果，複比と文字形状情\n報を併用することで，射影変換歪みを受けても非常に高い認識精度が得られることがわかった．","en":"In this paper, we propose a data embedding method for camera-based \ncharacter recognition as easy and accurate as bar-code reading.\nSpecifically, a horizontal stripe pattern, called cross ratio pattern, \nis embedded into a font image of a character category. Since the cross \nratio obtained from the pattern is invariant to projective distortion, \nit always provides correct category candidates regardless of camera angle.\nExperimental results showed that (i) the cross ratio can be extracted from \ndistorted character images with very high accuracy and thus (ii) the \ncross ratio can be used for enhancing recognition performance \nof conventional matching-based recognizers."}},{"key":"Uchida2006_IWFHR","id":"942","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Seiichi Uchida","en":""},{"ja":"Kazuhiro Tanaka","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"A Data-Embedding Pen"},"journal":{"ja":"Proc. 10th International Workshop on Frontiers in Handwriting Recognition (IWFHR2006)"},"num_pages":"6","year":2006,"month":10,"location":"La Baule, France","review":"reviewed","fields":["pattern-recognition"],"abstract":{"ja":"In order to use handwriting as a universal manmachine\ninterface, we assume a special pen device, called\ndata-embedding pen, which can embed binary data into\nhandwriting as a sequence of invisible ink drops along the\nhandwriting in a real-time manner. This paper describes\nthe assumed hardware, applications, and required technologies\nof the data-embedding pen. Especially, an accurate\nstroke recovery algorithm is proposed for retrieving\nthe data embedded in drawing order. In the algorithm, the\nembedded data itself is fully utilized to improve the accuracy\nof the recovery. A simulation experiment showed\nthat the algorithm can attain high accuracy on the stroke\nrecovery and the data retrieval."}},{"key":"Utsumi2006_RSJ","id":"1288","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"","en":"Yuzuko Utsumi"},{"ja":"","en":"Yoshio Iwai"},{"ja":"","en":"Masahiko Yachida"}],"title":{"en":"Performance evaluation of face recognition in the wavelet domain"},"journal":{"en":"Proceedings on IEEE/RSJ International Conference on Intelligent Robots and System"},"pages":"3344-3351","year":2006,"month":10,"location":"Beijing, China","review":"reviewed","fields":[]},{"key":"Kise2006_ICPR","id":"841","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Tomohiro Nakai","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Satoshi Yokota","en":""}],"title":{"ja":"Efficient Recognition of Planar Objects Based on Hashing of Keypoints — An Approach Towards Making the Physical World Clickable"},"journal":{"ja":"Proc. of the 18th International Conference of Pattern Recognition (ICPR2006)"},"volume":"4","pages":"813-816","year":2006,"month":8,"location":"Hong Kong","review":"reviewed","fields":["document-image-retrieval","document-image-analysis"],"doi":"10.1109/ICPR.2006.480","abstract":{"ja":"This paper presents a method of planar object recognition\nfor aiming at accessing information about objects by\ntaking pictures of them. For this purpose efficiency of processing\nis the central issue because current state-of-the-art\ntechnologies with tree structures do not necessarily work\nwell with a large amount of data represented as high dimensional\nvectors. To solve this problem, we employ hashing of\nkeypoints extracted from images of objects. With the help\nof hash keys obtained as integers converted from the real\nvalued vectors, keypoints are stored with object IDs and\nretrieved with no search process. Voting for object IDs is\nemployed to determine a recognized object as the one with\nthe largest vote. Experimental results show that the proposed\nmethod is at least 400 times faster than a brute-force\nmethod while 90% of objects were correctly recognized."}},{"key":"Omachi2006_ICPR","id":"839","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Shinichiro Omachi","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Seiichi Uchida","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Affine Invariant Information Embedment for Accurate Camera-Based Character Recognition"},"journal":{"ja":"Proc. of the 18th International Conference of Pattern Recognition (ICPR2006)"},"volume":"2","pages":"1098-1101","year":2006,"month":8,"location":"Hong Kong","review":"reviewed","fields":["pattern-recognition"],"doi":"10.1109/ICPR.2006.229","abstract":{"ja":"Recognizing characters in a scene image taken by a digital\ncamera has been studied for decades. However, it is still\na challenging problem to achieve high accuracy. In this paper,\nwe propose a method of embedding information in a\ncharacter pattern so that the class of the character can be\nidentified. The information should be robust against geometric\ndistortions since an image taken by a digital camera\nis usually geometrically distorted. In the proposed method,\na character pattern is designed in two colors so that the information\nis embedded as the area ratio of regions of two\ncolors. Since the area ratio is affine invariant, it is expected\nthat the area ratio is correctly extracted even if a character\nimage is affine-transformed. We generate character patterns\nwith the embedded information and discuss the effectiveness\nof the proposed method."}},{"key":"Uchida2006_ICPR","id":"840","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Seiichi Uchida","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"OCR Fonts Revisited for Camera-Based Character Recognition"},"journal":{"ja":"Proc. of the 18th International Conference of Pattern Recognition (ICPR2006)"},"volume":"2","pages":"1134-1137","year":2006,"month":8,"location":"Hong Kong","review":"reviewed","fields":["pattern-recognition"],"doi":"10.1109/ICPR.2006.891","abstract":{"ja":"In order to realize accurate camera-based character\nrecognition, machine-readable class information is embedded\ninto each character image. Specifically, each character\nimage is printed with a pattern which comprises five\nstripes and the cross ratio derived from the pattern represents\nclass information. Since the cross ratio is a projective\ninvariant, the class information is extracted correctly\nregardless of camera angle. The results of simulation experiments\nshowed that recognition rates over 99% were obtained\nby the extracted cross ratio under heavy projective\ndistortions."}},{"key":"Nakai2006_DAS","id":"383","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Tomohiro Nakai","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Use of Affine Invariants in Locally Likely Arrangement Hashing for Camera-Based Document Image Retrieval"},"journal":{"ja":"Lecture Notes in Computer Science (7th International Workshop DAS2006)"},"volume":"3872","pages":"541-552","year":2006,"month":2,"location":"Nelson, New Zealand","review":"reviewed","fields":["document-image-analysis","document-image-retrieval"],"doi":"10.1007/11669487_48","abstract":{"ja":"Camera-based document image retrieval is a task of searching document images from the database based on query images captured using digital cameras. For this task, it is required to solve the problem of “perspective distortion”of images, as well as to establish a way of matching document images efficiently. To solve these problems we have proposed a method called Locally Likely Arrangement Hashing (LLAH) which is characterized by both the use of a perspective invariant to cope with the distortion and the efficiency: LLAH only requires O(N) time where N is the number of feature points that describe the query image. In this paper, we introduce into LLAH an affine invariant instead of the perspective invariant so as to improve its adjustability. Experimental results show that the use of the affine invariant enables us to improve either the accuracy from 96.2% to 97.8%, or the retrieval time from 112 msec./query to 75 msec./query by selecting parameters of processing."}},{"key":"Nakai2006_VEV","id":"964","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"多数の特徴点によって記述されるパターンの高速認識法","en":"A Method of Fast Recognition of Patterns Represented by Many Feature Points"},"journal":{"ja":"ViEW2006講演論文集","en":"Proc. Vision Engineering Workshop ViEW2006"},"pages":"249-254","year":2006,"month":12,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"],"abstract":{"ja":"パターン認識の手法として，パターンから特徴点を抽出し，特徴点の配置に基づいて認識するものが提案されている．しかし，従来手法には，パターンに含まれる特徴点の数が多いと処理時間が膨大になるという問題がある．我々はすでに，特徴点の局所的配置を利用することで多数の特徴点で記述されるパターンを高速に認識する手法を提案している．また，この手法の高速性を利用した拡張として，英文文書を対象としたリアルタイム文書画像検索法についても提案している．本稿では，このパターン認識法の実用性を確認するために，様々な条件でWebカメラを用いたリアルタイム文書画像検索の実験を行った．その結果，認識対象の一部からでもほぼ100%の検索が可能であり，低い解像度や紙面の湾曲に対するロバスト性をもつことが示された．"}},{"key":"Amano2006","id":"960","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"天野 敏之","en":"Toshiyuki AMANO"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"},{"ja":"岡部 孝弘","en":"Takahiro OKABE"},{"ja":"加藤 毅","en":"Tsuyoshi KATO"},{"ja":"玉木 徹","en":"Toru TAMAKI"},{"ja":"若手プログラム参加者","en":"Participants in MIRU Young Researchers’ Program 2006"},{"ja":"内田 誠一","en":"Seiichi UCHIDA"}],"title":{"ja":"MIRU2006若手プログラム報告","en":"A Report on MIRU2006 Young Researchers’ Program"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"106","number":"429","presen_id":"PRMU2006-183","pages":"73-84","year":2006,"month":12,"review":"not_reviewed","fields":[],"abstract":{"ja":"2006 年7 月，仙台市青年文化センターにて画像の認識・理解シンポジウムMIRU2006 が開催された．そ\nの併設イベントとして，若手研究者の活性化および相互交流を図ることを目的とした「若手プログラム― 突撃イン\nタビュー！先人の成功と失敗に学ぼう」を開催した．本稿ではまず企画概要を述べた後，若手研究者が先輩研究者に\n対して行ったインタビューについてダイジェスト形式で報告する．","en":"MIRU Young Researchers’ Program was held in conjunction with MIRU2006 symposium for promoting\nexchanges between young researchers in CV and PR. This report gives a summary of the program."}},{"key":"Iwamura2006_IEICEPRMU_Better","id":"961","type":"domestic_conf","language":"EN","show":"show","authors":[{"ja":"","en":"Masakazu IWAMURA"},{"ja":"","en":"Yoshio FURUYA"},{"ja":"","en":"Koichi KISE"},{"ja":"","en":"Shinichiro OMACHI"},{"ja":"","en":"Seiichi UCHIDA"}],"title":{"en":"Better Decision Boundary for Pattern Recognition with Supplementary Information"},"journal":{"en":"IEICE Technical Report"},"volume":"106","number":"PRMU-376","presen_id":"PRMU2006-156","pages":"159-164","year":2006,"month":11,"review":"not_reviewed","fields":["pattern-recognition"],"note":{"ja":"KJPR2006."},"abstract":{"en":"Pattern recognition with supplementary information which differs from the conventional pattern recognition\nhas been proposed. This framework decreases misrecognition rates by using not only a pattern itself but\nalso supplementary information that assists recognition. In the framework, a classifier suitable for the conventional\npattern recognition has simply been used. Thus, there is room for improvement in the classifier. In this research, in\norder to improve the recognition performance with supplementary information, we attempt to change the recognition\nresults of the classifier by the shift of decision boundary from the Bayesian one. Though this reduces recognition\nrates in the conventional pattern recognition, this increases those in the pattern recognition with supplementary\ninformation. We have confirmed this by recognition experiments."}},{"key":"Furuya2006_IEICEPRMU","id":"959","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"古谷 嘉男","en":"Yoshio Furuya"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"内田 誠一","en":"Seiichi Uchida"}],"title":{"ja":"付加情報を用いるパターン認識における識別境界移動の効果—正規分布に従う人工サンプルを用いた場合の実験的検討—","en":"Effect of Moving Decision Boundaries in Pattern Recognition with Supplementary Information—Experimental Examination of Using Quadratic Discriminant Function—"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"106","number":"PRMU-301","presen_id":"PRMU2006-113","pages":"53-38","year":2006,"month":10,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"通常のパターン認識とは異なる，付加情報を用いるパターン認識という新しい枠組みが提案されている．これは，通常のパターンと同時にそれを補助する情報(付加情報)をパターンとは別に用意し，それらを同時に用いることにより誤認識率を減少させる方式である．この方式において，現在提案されている方法では通常のパターン認識で誤認識率が最小になるように設計された識別器をそのまま用いているが，付加情報を用いるパターン認識においては必ずしもそのような識別器が最良ではないと考えられる．そこで，本研究では識別器の識別境界を，誤認識率を最小にするベイズ識別境界\nから移動することを試みる．この変更は付加情報を用いない従来のパターン認識における誤認識率を増加させるが，付加情報を用いるパターン認識における誤認識率を減少させる．このことを実験により確認した．","en":"Pattern recognition with supplementary information which differs from the conventional pattern recognition\nhas been proposed. This framework is capable of decreasing error rates by using not only a pattern itself but\nalso its supplementary information that assists recognition. In the previous report, we confirmed a better recognition\nrate is achievable by the shift of the decision boundaries from the Bayesian ones in the experiment using a character\ndata set. However, it was not the strict proof of the existence of the achievability because the Bayesian decision\nboundaries are estimates. In this report, to confirm the achievability in the strict sense, we make use of the artificial\nsamples following the normal distributions. This enables us to obtain the Bayesian decision boundaries precisely."}},{"key":"Iwamura2006_IEICEPRMU_Accuracy","id":"945","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"特徴点の局所的配置に基づく物体認識の精度・速度に関する一考察","en":"On Accuracy and Speed of Object Recognition Based on Local Arrangements of Feature Points"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"106","number":"PRMU-229","presen_id":"PRMU2006-67","pages":"49-56","year":2006,"month":9,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"],"note":{"ja":"PPTファイルを開くとパスワードを尋ねられますので、「読み取り専用」で開いてください。\n<br>\nCVIM155-8."},"abstract":{"ja":"我々は既に，特徴点の局所的配置に基づく高精度かつ高速な文書画像検索方法\n(LLAH)を提案している．これは，広く知られた特徴点ベースの物体認識手法で\nあるGeometric Hashingを精度・速度の両面において著しく改善する手法である．\n本報告ではこの手法をGeometric Hashingと比較することにより，どのような\n要因が高精度化と高速化に寄与しているかを考察する．\nまた，LLAHの検索性能は検索対象の撮影角度に依存するため，\n撮影角度とLLAHの検索性能の関係についても考察する．","en":"The geometric hashing is a well-known object recognition technique\nbased on the arrangements of feature points.\nWe have proposed ``locally likely arrangement hashing (LLAH)''\nwhich outperforms the geometric hashing\nin both retrieval accuracy and processing time.\nIn this report, by comparing both methods,\nwe consider the major factors which bring the improvement.\nWe also consider the relationship between \na picture angle and the accuracy of the LLAH\nbecause the accuracy of the LLAH depends on the picture angle."}},{"key":"Nakai2006_IEICEPRMU_Real","id":"955","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"特徴点の局所的配置に基づくリアルタイム文書画像検索とその拡張現実への応用","en":"Real-Time Document Image Retrieval Based on Local Arrangements of Feature Points and Its Application to Augmented Reality"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"106","number":"PRMU-229","presen_id":"PRMU2006-66","pages":"41-48","year":2006,"month":9,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"],"awards":{"ja":["平成19年度山下記念研究賞受賞"],"en":["2007 IPSJ Yamashita SIG Research Award Winner"]},"note":{"ja":"CVIM155-7"},"abstract":{"ja":"本稿で示すリアルタイム文書画像検索とは，\nWebカメラを用いて撮影された文書に対応する電子文書を\nデータベースからリアルタイムで検索する処理である．\n我々はすでに特徴点の局所的配置に基づく文書画像検索法を提案しており，\nリアルタイム文書画像検索はこの手法の高速性を利用した\nリアルタイム処理への応用である．\nまた．この手法では検索の過程で文書の部分ごとの対応関係が得られるため，\n撮影範囲や紙面の傾きを推定することができる．\nそこで，リアルタイム文書画像検索の応用として，\n文書の関連情報を撮影画像に重畳表示する拡張現実も示す．","en":"In this report, we propose a method of real-time document image retrieval.\nIt is a task of searching document images from the database\nbased on query images captured by a web-camera.\nWe have already proposed a method of document image retrieval based on\nlocal arrangements of feature points.\nReal-time document image retrieval is realized by applying the method to\neach frame captured by a web-camera.\nAs an application of the real-time retrieval, we also propose\na method of augmented reality for documents. This method is to\nsuperimpose relevant information onto the camera-captured image\nnaturally by using the parameters of perspective transformation\ncalculated as a subsidiary result of retrieval."}},{"key":"Noguchi2006","id":"954","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"野口 和人","en":"Kazuto Noguchi"},{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"特徴ベクトルの近傍探索と物体認識の効率に関する実験的検討","en":"Experimental Investigation of Relation Between Near Neighbor Search Methods for Feature Vectors and Efficiency of Object Recognition"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"106","number":"PRMU-229","presen_id":"PRMU2006-68","pages":"57-64","year":2006,"month":9,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval","object-recognition"],"note":{"ja":"CVIM155-9"},"abstract":{"ja":"画像の局所記述子としてSIFTやPCA-SIFTを用いる物体認識では，\n一画像あたりの特徴ベクトルの数が膨大となるため，\n特徴ベクトルの照合速度が全体の効率を左右する．\n特徴ベクトルの照合は，最近傍探索の枠組みで捉えられるため，\n最近傍探索の高速化が中心的課題となる．本稿では，\n「物体認識のためには，正しく照合される特徴ベクトルの数が，\n誤って他の物体に照合される数を上回ればよい」という観点から，\n近似最近傍探索における距離計算の回数を大幅に削減する2つの手法を提案する．\n一つは，多くの距離計算が避けられない特徴ベクトルを\n削除する方法である．もう一つは，全く距離計算を行わない手法である．\n1万画像のデータベース，2,000画像の検索質問を用いた実験から，\nANNを用いる手法と比べて同じ認識率を実現するために必要な処理時間を1/2から1/3に削減でき，\n例えば認識率98%，処理時間8.3ms/queryを達成できることが分かった．","en":"Efficiency of object recognition methods using local descriptors such as SIFT and\nPCA-SIFT depends largely on the speed of matching between feature\nvectors since images are described by a large number of feature vectors.\nBecause the matching is considered to be ``nearest neighbor (NN) search''\nof feature vectors, the problem is paraphrased by ``how to make \nthe NN search efficient''. For the object recognition, it is required\nthat the number of incorrect matching does not exceed that of correct\nmatching. In other words, a certain number of incorrect matching is\nacceptable. This observation allows us to make NN search more efficient\nusing approximate NN search with reduced distance calculation.\nFor this purpose,  we propose two methods: one is to eliminate \nfeature vectors that require a number of distance calculations. \nThe other is to use no distance calculation.\nFrom experimental results with 10,000 database images and 2,000 query images,\nit is shown that the proposed method is two to three times\nefficient as compared to a method using ANN and can achieve,\nrecognition rate of 98% with 8.3 ms/query."}},{"key":"Utsumi2006_MIRU","id":"1296","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"内海ゆづ子","en":""},{"ja":"岩井儀雄","en":""},{"ja":"谷内田正彦","en":""}],"title":{"ja":"顔認識のためのウェーブレット特徴量の評価"},"journal":{"ja":"画像の認識・理解シンポジウム（MIRU2006)"},"pages":"1302-1307 No. IS-3-49","year":2006,"month":7,"review":"not_reviewed","fields":[]},{"key":"Furuya2006_MIRU","id":"843","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"古谷 嘉男","en":"Yoshio Furuya"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"内田 誠一","en":"Seiichi Uchida"}],"title":{"ja":"付加情報を用いるパターン認識における識別境界と誤認識率の関係","en":"Relationship Between Decision Boundaries and Recognition Error Rates in Pattern Recognition with Supplementary Information"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2006)論文集","en":"Proceedings of MIRU 2006"},"presen_id":"IS3-37","pages":"1231-1236","year":2006,"month":7,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"通常のパターン認識とは異なる，付加情報を用いるパターン認識という新しい枠組みが提案されている．こ\nれは，通常のパターンと同時にそれを補助する情報(付加情報) をパターンとは別に用意し，それらを同時に用いるこ\nとにより誤認識率を減少させる方式である．この方式において，現在提案されている方法では通常のパターン認識で\n誤認識率が最小になるように設計された識別器をそのまま用いているが，付加情報を用いるパターン認識においては\n必ずしもそのような識別器が最良ではないと考えられる．そこで，本研究では識別器の識別境界を，誤認識率を最小に\nするベイズ識別境界から移動することを試みる．この変更は付加情報を用いない従来のパターン認識における誤認識\n率を増加させるが，付加情報を用いるパターン認識における誤認識率を減少させる．このことを実験により確認した．","en":"Pattern recognition with supplementary information which differs from the conventional pattern recognition\nhas been proposed. This framework decreases misrecognition rates by using not only a pattern itself but\nalso supplementary information that assists recognition. In the framework, a classifier suitable for the conventional\npattern recognition has simply been used, and thus there is room for improvement in the classifier. In this research,\nin order to improve the recognition performance with supplementary information, we try to change the recognition\nresults of the classifier by move of decision boundaries from the Bayesian ones. Though this reduces recognition\nrates in the conventional pattern recognition, this increases those in the pattern recognition with supplementary\ninformation. We have confirmed this by recognition experiments."}},{"key":"Iwamura2006_MIRU_Character","id":"842","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"根岸 和也","en":"Kazuya Negishi"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"阿曽 弘具","en":"Hirotomo Aso"}],"title":{"ja":"変形を許容する高速なテンプレートマッチングによる情景画像中文字認識","en":"Character Recognition in Scene Images by Fast Deformable Template Matching"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2006)論文集","en":"Proceedings of MIRU 2006"},"presen_id":"IS2-42","pages":"906-911","year":2006,"month":7,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"情景中の文字を認識するためには，一般に前処理として文字切り出しが必要であるが，既存の手法では文\n字列を成さない「孤立文字」や隣接した2 文字が接触した接触文字などを切り出せないという問題がある．また，情\n景画像中の文字画像は射影変換などの変形を受けることも知られている．そこで本論文では，射影変換や字形の違い\nに対する頑健さを持ち，孤立文字や接触文字など，既存の手法では切り出せない文字の切り出しが可能で，かつ高速\nな手法を提案する．提案手法を実際の画像に適用したところ，射影変換や字形の違いに対して頑健であり，孤立文字\nである時計の文字盤等や，数式中の接触文字を良好に，また単純なテンプレートマッチング手法より高速に切り出せ\nることを確認できた．","en":"To recognize characters in scene images, a character segmentation method is generally required for\npre-processing. However, there are problems that traditional methods are not able to segment “isolated character”\nwhich does not constitute a character string and “touching character” which is connected to other characters.\nMoreover, it is known that character images in scene images suffer from perspective distortion. In this paper, we\npropose a new fast character segmentation method of characters including the isolated and the touching ones with\nrobustness for perspective distortion and character deformations. By applying the proposed method to scene images,\nit is confirmed that the method is robust against perspective distortion and character deformations, and the method\nis superior in computational time against a simple template matching method."}},{"key":"Nakai2006_MIRU","id":"844","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"Webカメラを用いたリアルタイム文書画像検索とその拡張現実への応用","en":"Real-Time Document Image Retrieval using a Web Camera and Its Application to Augmented Reality"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2006)論文集","en":"Proceedings of MIRU 2006"},"presen_id":"DS-5","pages":"1379-1380","year":2006,"month":7,"review":"not_reviewed","fields":["document-image-retrieval","document-image-analysis"],"awards":{"ja":["MIRU2006 デモセッション優秀賞受賞"],"en":["MIRU2006 Best Demonstration Award Winner"]},"abstract":{"ja":"本発表で示すリアルタイム文書画像検索とは，Webカメラを用いて撮影した文書に対応する文書画像をデータベースからリアルタイムで検索する処理である．我々はすでにデジタルカメラを用いた文書画像検索法を提案しており，本発表はこの手法の高速性を利用したリアルタイム処理への応用である．また，この手法では検索の過程で文書の部分ごとの対応関係が得られるため，紙面の傾きと撮影された領域を推定することができる．これらの情報を利用することで，文書の関連情報を撮影画像に重ねて表示する拡張現実への応用が可能になる．"}},{"key":"Omachi2006_MIRU","id":"845","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"面積比による付加情報を用いた実環境文字認識","en":"Camera-Based Character Recognition with Supplementary Informationby Area Ratio"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2006)論文集","en":"Proceedings of MIRU 2006"},"presen_id":"OS5A-2","pages":"187-192","year":2006,"month":7,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"ディジタルカメラを入力デバイスとして環境中の文字を高精度に認識するために，文字画像と同時に認識\n補助のための付加情報を提示する方法が検討されている．付加情報は，人間にとって自然な形で提示されること，お\nよび，幾何学的変形に対してロバストに抽出できることが要求される．本論文では，これらの要求を満たす手法とし\nて，面積比を利用した付加情報提示手法を提案する．すなわち，文字パターンを2 色で印字することを前提とし，そ\nれぞれの色の領域の面積比を特定の値とするようにデザインする．面積比はアフィン変換に不変であり，アフィン変\n換を受けた環境においても誤りなく抽出されることが期待される．実際に付加情報を埋め込んだ文字パターンを作成\nし，ディジタルカメラで撮影された画像中の文字を認識する実験を行い，提案手法の有効性を確認する．","en":"In order to achieve a highly accurate recognition of characters in a scene image with a digital camera,\nthere are some attempts on offering supplementary information for recognition with a character image. The information\nshould be human-friendly and robust against geometric distortions. In this paper, we propose a method\nof embedding information in a character pattern by designing a character pattern in two colors so that the information\nis embedded as the area ratio of the two colors. It is expected that the area ratio is correctly extracted\neven if the character pattern is affine-transformed since the area ratio is affine invariant. The effectiveness of the\nproposed method is confirmed by generating character patterns with the embedded information and carrying on an\nexperiment of recognizing characters in a scene image taken by a digital camera."}},{"key":"Sakai2006","id":"846","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"酒井 恵","en":"Megumi Sakai"},{"ja":"内田 誠一","en":"Seiichi Uchida"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"実環境文字認識のためのオートマトン制御付き最適境界検出法","en":"FSA-Based Optimal Segmentation for Camera-Based CharacterRecognition"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2006)論文集","en":"Proceedings of MIRU 2006"},"presen_id":"IS3-2","pages":"1024-1029","year":2006,"month":7,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"カメラを用いた実環境中の文字パターンの高精度認識を目標として，複比パターンと呼ばれる縞模様状の\nカテゴリ情報を文字パターンそのものに埋め込む方法が検討されている．この方法では，各縞の幅から計算される複\n比の値が文字カテゴリを表している．従ってカテゴリ情報抽出のためには，各縞の境界を検出する必要がある．本論\n文では，動的計画法と有限状態オートマトン(FSA) を組み合わせたロバストな境界検出法を提案する．評価実験の結\n果，本手法により安定して境界を検出できることがわかった．さらに検出された境界から計算した複比の値を用いて\n文字認識を行なったところ，非常に高い認識率が得られた．","en":"For accurate camera-based character recognition, a method where category information is embedded\nonto every character image as a striped pattern has been investigated. The cross ratio calculated from the widths of\nstripes represents the category of the character. In this paper, an algorithm to detect the boundaries of the stripes\nfor measuring the widths is proposed. The algorithm is realized by a combination of a dynamic programming-based\noptimal segmentation framework and a finite state automaton (FSA) which represents the characteristics of the\nstriped pattern. Experimental results showed the ability of the proposed method on not only accurate boundary\ndetection but also accurate recovery of the embedded cross ratio."}},{"key":"Tanaka2006_MIRU","id":"943","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"田中 一弘","en":"Kazuhiro TANAKA"},{"ja":"内田 誠一","en":"Seiichi UCHIDA"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"},{"ja":"大町 真一郎","en":"Shinichiro OMACHI"},{"ja":"黄瀬 浩一","en":"Koichi KISE"}],"title":{"ja":"データ埋め込みペンによる機械可読な手書きパターン","en":"Machine-Readable Handwriting Patterns by Data-Embedding Pen"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU2006)論文集","en":"Proceedings of MIRU 2006"},"presen_id":"IS3-2","pages":"1296-1301","year":2006,"month":7,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"本論文では，紙とペンによる手書きパターンを情報埋め込み技術により機能拡張する方法について述べる．\n具体的には，機械可読な情報を手書きパターンにリアルタイムに埋め込む機能を具備したペンを考え，これにより手\n書きパターンを，タグ，バーコード，サイバー空間とのリンクとして利用することを考える．本論文では，ペンのハー\nドウエア仕様，ならびに情報埋め込み・抽出手段，応用先について述べる．また特に，情報抽出の際の中心的問題に\nなる筆順復元について詳述する．","en":"In order to use handwritings as a universal man-machine interface, we assume a pen device ―\ndata-embedding pen ― which can embed digital data into a handwriting by invisible ink in a real-time manner.\nThis paper discusses the system design, application, and required technologies around the data-embedding\npen. Especially, a novel stroke recovery algorithm is proposed for retrieving the embedded data along writing order.\nIn the algorithm, embedded data is used to help the recovery. A simulation experiment showed that the algorithm\ncan attain high accuracy on the stroke recovery and the data retrieval."}},{"key":"Iwamura2006_MIRU_Position","id":"946","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"阿曽 弘具","en":"Hirotomo Aso"}],"title":{"ja":"部分空間法における原点の位置と認識性能に関する考察","en":"On the Position of the Origin and Recognition Performance of Subspace Method"},"journal":{"ja":"部分空間法研究会2006予稿集","en":"Proc. Subspace2006 in conjunction with MIRU2006"},"pages":"42-51","year":2006,"month":7,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"部分空間法は，簡潔で，かつ望ましい性質を持つため，パターン認識においてよく用いられる認識手法の\n一つである．様々な改良法が提案されているが，そのほとんど全ては「暗に」特徴ベクトルの原点に依存している．\nところが，原点の位置が部分空間法の認識性能にどのような影響を与えるのか，十分に解明されていない．本稿では，\n「コアアフィン部分空間」と「誤認識穴」という概念を導入し，この問題について議論する．また，よく知られた部分\n空間法であるCLAFIC 法と投影距離法の，ある種の類似性についても言及する．","en":"Subspace method is one of the most popular recognition methods in pattern recognition because of its\nsimplicity and some desirable properties. Many varieties of the subspace methods have been proposed. Though most\nof them implicitly depend on the origin of feature vectors, the effect of the position of the origin to the recognition\nperformance have not analyzed enough. In this paper, we discuss and address this problem by introducing the\nconcepts of core affine subspace and misclassification hole. We also mention the similarity between the CLAFIC\nmethod and the projection distance in a sense under an assumption."}},{"key":"Negishi2006","id":"392","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"根岸 和也","en":"Kazuya Negishi"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"阿曽 弘具","en":"Hirotomo Aso"}],"title":{"ja":"情景画像中文字認識のための変形を許容する高速なテンプレートマッチング","en":"Fast Deformable Template Matching for Character Recognition in Scene Images"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"105","number":"PRMU-673","presen_id":"PRMU2005-247","pages":"101-106","year":2006,"month":3,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"情景中の文字を認識するためには，一般に前処理として文字切り出しが\n必要であるが，既存の手法では文字列を成さない「孤立文字」や隣接し\nた2文字が接触した接触文字などを切り出せないという問題がある．\nまた，情景画像中の文字画像は射影変換などの変形を受けることも知られてい\nる．\nそこで本論文では，射影変換や字形の違いに対する頑健さを持ち，孤立文字や\n接触文字など，既存の手法では切り出せない文字の切り出しが可能でかつ高速な\n手法を提案する．\n提案手法を実際の画像に適用したところ，\n射影変換や字形の違いに対して頑健であり，孤立文字である時計の文字盤等や，\n数式中の接触文字を良好に，また単純なテンプレートマッチング手法より高速に切り出せることを確認した．","en":"To recognize characters in scene images, a character segmentation method is generally required for pre-processing. \nHowever, there are problems that traditional methods are not able to segment ``isolated character''\nwhich does not constitute a character string and ``touching character'' which is connected to other characters. \nMoreover, it is known that character images in scene images suffer from perspective distortion. \nIn this paper, we propose a new fast character segmentation method of\ncharacters including the isolated and the touching ones with\nrobustness for perspective distortion and character deformations.\nBy applying the proposed method to scene images, \nit is confirmed that the method is robust against perspective distortion and character\ndeformations, and the method is \nsuperior in computational time against a simple template matching method."}},{"key":"Nakai2006_IEICEPRMU_Use","id":"408","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"デジタルカメラを用いた高速文書画像検索におけるアフィン不変量および相似不変量の利用","en":"Use of Affine Invariants and Similarity Invariants in Fast Camera-based Document Image Retrieval"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"Technical Report of IEICE"},"volume":"105","number":"PRMU-614","presen_id":"PRMU2005-188","pages":"25-30","year":2006,"month":2,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"],"abstract":{"ja":"我々はすでにデジタルカメラを用いた高速文書画像検索の一手法として特徴点の局所的配置に基づく手法を提案している．この手法では射影不変量である複比を用いているが，複比は特徴点の位置の変動に対して不安定であることが問題となっている．そこで，本稿ではより安定な性質をもつアフィン不変量および相似不変量の利用を考える．実験により，アフィン不変量を用いることで画像の歪みが大きい場合でも高精度な検索が可能であることが示された．また，相似不変量は，画像の歪みに対してロバストでないため，大きな射影歪みの生じる場合には適さないことが示された．","en":"We have already proposed a method of camera-based document image retrieval which utilizes perspective invariants called cross-ratios. However, the cross-ratios have defects that they are unstable against changes of positions of feature points. In this paper, we evaluate the use of affine invariants and similarity invariants which are considered to be more stable. Experimental results have shown that higher accuracy is obtained with affine invarinats even under perspective distortion. It is also shown that similarity invariants are not suitable for retrieval of largely distorted images since they are not robust to perspective distortion."}},{"key":"Omachi2006_IEICEPRMU","id":"384","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"大町 真一郎","en":"Shinichiro OMACHI"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"},{"ja":"内田 誠一","en":"Seiichi UCHIDA"},{"ja":"黄瀬 浩一","en":"Koichi KISE"}],"title":{"ja":"高精度実環境文字認識のための面積比による付加情報埋め込み","en":"Area Ratio as Supplementary Information for Camera-Based Character Recognition"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"105","number":"PRMU-615","presen_id":"PRMU2005-213","pages":"67-72","year":2006,"month":2,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"環境中の文字をディジタルカメラを入力デバイスとして高精度に認識するために，文字画像と同時に認識\n補助のための付加情報を提示する方法が検討されている．付加情報は，幾何学的変換に対してロバストに抽出できる\nことが要求される．本報告では，面積比を利用した付加情報提示手法を提案する．面積比はアフィン変換に不変であ\nり，アフィン変換を受けた環境においても誤りなく抽出されることが期待される．実際に付加情報を埋め込んだ文字\nパターンを作成し，提案手法の有効性と可能性について議論する．","en":"In order to achieve a highly accurate recognition of characters in a scene image with a digital camera,\nthere are some attempts on offering supplementary information for recognition with a character image. The\ninformation should be robust against geometric distortions since an image taken by a digital camera is usually\ngeometrically distorted. In this paper, we propose a method of embedding information in a character pattern by\ndesigning a character pattern in two colors so that the information is embedded as the area ratio of the two colors.\nIt is expected that the area ratio is correctly extracted even if the character pattern is affine-transformed since the\narea ratio is affine invariant. We evaluate generated character patterns with the embedded information and discuss\nthe effectiveness of the proposed method."}},{"key":"Tanaka2006_IEICEPRMU","id":"385","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"田中 一弘","en":"Kazuhiro TANAKA"},{"ja":"内田 誠一","en":"Seiichi UCHIDA"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"},{"ja":"大町 真一郎","en":"Shinichiro OMACHI"},{"ja":"黄瀬 浩一","en":"Koichi KISE"}],"title":{"ja":"手書きパターンへの情報埋め込みと情報抽出","en":"On-line data embedding into handwriting patterns"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"105","number":"PRMU-615","presen_id":"PRMU2005-212","pages":"61-66","year":2006,"month":2,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"本研究では，情報埋め込み技術を用いることで，紙とペンで生成される手書きパターンの価値を高めるこ\nとを目標とする．本報告ではそのために想定しているハードウエア，ならびに情報埋め込み・抽出手段，応用先につ\nいて述べる．さらに，情報抽出の際の中心的問題になる筆順復元について検討する．この筆順復元については，情報\nが埋め込みされているという本手書きパターン特有の状況を活用することで，高精度化が可能であることを示す．","en":"In order to use handwritings as a universal man-machine interface, \nwe assume a pen device ― data-embedding pen ― which can \nembed digital data into a handwriting by invisible ink in a real-time manner.\nThis paper discusses the system design, application, and required technologies \naround the data-embedding pen. Especially, a novel stroke recovery \nalgorithm is proposed for retrieving the embedded data along writing\n order. In the algorithm, embedded data is used to help the recovery. \nA simulation experiment showed that the algorithm can attain high \naccuracy on the stroke recovery and the data retrieval."}},{"key":"Utsumi2006_PRMU","id":"1292","type":"domestic_conf","language":"JA","show":"show_except_datafile","authors":[{"ja":"内海ゆづ子","en":""},{"ja":"岩井儀雄","en":""},{"ja":"谷内田正彦","en":""}],"title":{"ja":"顔認識のためのウェーブレット特徴量の評価"},"presen_id":"PRMU, No. 2005-173","pages":"141-146,","year":2006,"month":1,"review":"not_reviewed","fields":[]},{"key":"Ohmae2006_SIGNL","id":"378","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"大前 信弘","en":"Nobuhiro OHMAE"},{"ja":"黄瀬 浩一","en":"Koichi KISE"}],"title":{"ja":"Webの表を対象とした属性の自動識別","en":"Automatic recognition of attributes from table in web pages"},"journal":{"ja":"情報処理学会研究報告","en":"SIG Technical Report"},"presen_id":"NL-171","pages":"43-48","year":2006,"month":1,"review":"not_reviewed","fields":["information-retrieval"]},{"key":"Takeuchi2006","id":"377","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"竹内 謹治","en":"Kinji Takeuchi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"類似画像とキーワードを利用したWeb画像の説明文抽出","en":"Extraction of Explanatory Text for Web Images Using Similar Images and Keywords"},"journal":{"ja":"情報処理学会研究報告","en":"SIG Technical Report"},"presen_id":"NL-171","pages":"7-12","year":2006,"month":1,"review":"not_reviewed","fields":["information-retrieval"]},{"key":"Klink2006","id":"1111","type":"book","language":"EN","show":"show","authors":[{"ja":"Stefan Klink","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"Markus Junker","en":""},{"ja":"Stefan Agne","en":""}],"title":{"ja":"Document Information Retrieval"},"book_title":"Digital Document Processing","pages":"351-378","year":2006,"month":12,"publisher":"Springer-Verlag","address":"London","review":"not_reviewed","fields":["document-image-analysis"]},{"key":"Nakai2006_MThesis","id":"402","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"中居 友弘","en":"Tomohiro Nakai"}],"title":{"ja":"特徴点の局所的配置に基づく高速文書画像検索"},"year":2006,"month":3,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"]},{"key":"Ohmae2006_MThesis","id":"401","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"大前信弘","en":""}],"title":{"ja":"Webの表構造を用いた属性の自動抽出"},"year":2006,"month":3,"review":"not_reviewed","fields":[]},{"key":"Furuya2006_BThesis","id":"398","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"古谷 嘉男","en":"Yoshio Furuya"}],"title":{"ja":"付加情報を用いるパターン認識における誤認識率の減少方法"},"year":2006,"month":3,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Kimura2006","id":"399","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"木村 綾","en":""}],"title":{"ja":"柔軟物体上の文字列・ロゴの自然な置き換え"},"year":2006,"month":3,"review":"not_reviewed","fields":[]},{"key":"Murakami2006","id":"394","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"村上 誠知","en":""}],"title":{"ja":"識別器カスケードを用いたシーン画像からの文字列抽出法"},"year":2006,"month":3,"review":"not_reviewed","fields":[]},{"key":"satoshi2006","id":"838","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"横田 哲","en":"yokota satoshi"}],"title":{"ja":"ハッシュに基づく平面物体の高速認識"},"year":2006,"month":3,"review":"not_reviewed","fields":["information-retrieval","pattern-recognition"]},{"key":"Sezaki2006","id":"396","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"瀬崎 直人","en":""}],"title":{"ja":"レイアウト情報と言語情報を用いた画像と説明文の対応付け"},"year":2006,"month":3,"review":"not_reviewed","fields":["information-retrieval"]},{"key":"Tanimoto2006","id":"395","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"谷本 真吾","en":""}],"title":{"ja":"HTMLの階層構造と言語情報を用いた手順記述の抽出"},"year":2006,"month":3,"review":"not_reviewed","fields":["information-retrieval","information-retrieval"]},{"key":"Kise2006_J","id":"956","type":"survey","language":"JA","show":"show_except_datafile","authors":[{"ja":"黄瀬 浩一","en":"Koichi KISE"},{"ja":"大町 真一郎","en":"Shinichiro OMACHI"},{"ja":"内田 誠一","en":"Seiichi UCHIDA"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"}],"title":{"ja":"ディジタルカメラによる文字・文書の認識・理解","en":"Recognition and Understanding of Characters and Documents Using Digital Cameras"},"journal":{"ja":"電子情報通信学会誌","en":"The Journal of the Institute of Electronics, Information and Communication Engineers"},"volume":"89","number":"9","pages":"836-841","year":2006,"month":9,"address":"東京","review":"not_reviewed","fields":["document-image-analysis","pattern-recognition"],"url":"https://www.journal.ieice.org/summary.php?id=k89_9_836","note":{"ja":"<a href=\"http://www.m.cs.osakafu-u.ac.jp/~masa/image/models.png\">図２のカラー版はこちら。</a>","en":"<a href=\"http://www.m.cs.osakafu-u.ac.jp/~masa/image/models.png\">The color version of Fig.2 is here.</a>"},"abstract":{"ja":"ディジタルスチルカメラやビデオカメラの普及と発展に伴って，撮影した画像内の文字・文書を情報処理に利用したいという要求が高まっている．本稿では，このようなカメラを用いた文字・文書の認識・理解を通して，我々は何を得ることができるのか，また実現には何が問題であり，現在どのような取組みがなされているのかについて解説する．加えて，残された研究課題について触れるとともに，ユーザインタフェースへの適用の視点から筆者らが進めている新しい試みについても紹介する．"}},{"key":"Kise2005_IEICE","id":"389","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Shota Fukushima","en":""},{"ja":"Keinosuke Matsumoto","en":""}],"title":{"ja":"Document Image Retrieval for QA Systems Based on the Density Distributions of Successive Terms"},"journal":{"ja":"IEICE Transactions on Information and Systems"},"volume":"E88-D","number":"8","pages":"1843-1851","year":2005,"month":8,"review":"reviewed","fields":["document-image-analysis"]},{"key":"Kise2005_IEEJ","id":"390","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"辻野 雅章","en":"Masaaki Tsujino"},{"ja":"松本 啓之亮","en":"Keinosuke Matsumoto"}],"title":{"ja":"文字の出現密度分布に基づく文書画像の関連部分検索","en":"Retrieval of Relevant Parts of Document Images Based on 2D Density Distributions of Characters"},"journal":{"ja":"電気学会論文誌（Ｃ）","en":"IEEJ Transactions on Electronics, Information and Systems"},"volume":"125","number":"1","pages":"113-119","year":2005,"month":1,"review":"reviewed","fields":["document-image-analysis"]},{"key":"Nakai2005_LI","id":"376","type":"letter","language":"JA","show":"show","authors":[{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"デジタルカメラによる文書画像検索—1万ページから0.1秒で検索する—","en":"Document Image Retrieval with Digital Cameras —Retrieval in 0.1 Second from 10,000 Pages—"},"journal":{"ja":"情報科学技術レターズ","en":"Information Technology Letters"},"volume":"4","presen_id":"LI-001","pages":"133-136","year":2005,"month":9,"review":"reviewed","fields":["document-image-retrieval"],"awards":{"ja":["FIT2005 ヤングリサーチャー賞受賞"],"en":["FIT2005 Young Researcher Award Winner"]}},{"key":"Uchida2005_MPR","id":"371","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Seiichi Uchida","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Design and recognition of human-readable and machine-readable patterns"},"journal":{"ja":"1st Joint Workshop on Machine Perception and Robotics (MPR2005)"},"num_pages":"8","year":2005,"month":9,"location":"Beijin, China","review":"reviewed","fields":["pattern-recognition","document-image-analysis"],"abstract":{"ja":"In this paper, design and recognition of human-readable\nand machine-readable patterns are investigated. Specifically\nspeaking, we design character images printed with a\nhorizontal stripe pattern, called a cross ratio pattern. The\ncross ratio derived from the cross ratio pattern represents\nthe class information of the character. Since the cross ratio\nis invariant to projective distortion, the class information is\nextracted correctly regardless of camera angle. The character\nimage itself is human-readable and therefore the character\nimage with the cross ratio pattern is not only humanreadable\nand but also machine-readable and can be used as\na medium for human-machine communication."}},{"key":"Iwamura2005_ICDAR","id":"365","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Kazuya Negishi","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Hirotomo Aso","en":""}],"title":{"ja":"Isolated Character Recognition by Searching Feature Points"},"journal":{"ja":"Proc. 8th International Conference on Document Analysis and Recognition (ICDAR2005)"},"pages":"1035-1039","year":2005,"month":9,"location":"Seoul, South Korea","review":"reviewed","fields":["pattern-recognition","document-image-analysis"],"doi":"10.1109/ICDAR.2005.147","abstract":{"ja":"Conventional segmentation technique cannot extract difficult characters\nsuch as an isolated character and a touching character.\nIn this paper, we propose a novel character recognition method which\nexecutes segmentation and recognition simultaneously.\nThis method enables us to extract and recognize such difficult\ncharacters.\nThe effectiveness of the proposed method is confirmed by experiments."}},{"key":"Nakai2005_ICDAR","id":"366","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Tomohiro Nakai","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Camera-Based Document Image Retrieval as Voting for Partial Signatures of Projective Invariants"},"journal":{"ja":"Proc. 8th International Conference on Document Analysis and Recognition (ICDAR2005)"},"pages":"379-383","year":2005,"month":9,"location":"Seoul, South Korea","review":"reviewed","fields":["document-image-analysis","document-image-retrieval"],"doi":"10.1109/ICDAR.2005.64","abstract":{"ja":"We propose a method of document image retrieval using digital cameras. The proposed method takes as input a part or the whole of a document acquired as a query by a digital camera, and retrieves a document image that includes the query. For this purpose, it is required to solve the problem of \"perspective distortion\" of images, as well as to establish a way of matching parts of document images flexibly. These are achieved based on the following characteristics of the proposed method: (1) Indexing of document images using the projective invariants called the \"cross-ratios\", (2) Retrieval as voting for partial signatures of document images defined by the cross-ratios. From experimental results using digital cameras with high and low resolutions, we demonstrate the effectiveness of the proposed method."}},{"key":"Iwamura2005_CBDAR","id":"367","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu IWAMURA","en":""},{"ja":"Seiichi UCHIDA","en":""},{"ja":"Shinichiro OMACHI","en":""},{"ja":"Koichi KISE","en":""}],"title":{"ja":"Recognition with Supplementary Information —How Many Bits Are Lacking for 100% Recognition?—"},"journal":{"ja":"Proc. First International Workshop on Camera-Based Document Analysis and Recognition (CBDAR2005)"},"pages":"68-75","year":2005,"month":8,"location":"Seoul, South Korea","review":"reviewed","fields":["pattern-recognition"],"abstract":{"ja":"In this paper, we propose a new model in which the classifier\nreceives not only a pattern itself but also supplementary\ninformation that assists recognition. This model enables\nus to achieve a 100% recognition rate with a 0%\nrejection rate with certain bits of supplementary information\nrequired. For printed characters, experiments show\nthat 4 bits of supplementary information were required in\nthe leave-one-out method and 1 bit was in the resubstitution\nmethod. In addition, we generalize the discussion into\nthe relationship among a quantity of supplementary information,\na recognition rate and a rejection rate. The theory\npresented in this paper is applied to the data embedding of\na font set for camera-based character recognition [9]."}},{"key":"Nakai2005_CBDAR","id":"369","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Tomohiro Nakai","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Hashing with Local Combinations of Feature Points and Its Application to Camera-Based Document Image Retrieval—Retrieval in 0.14 Second from 10,000 Pages—"},"journal":{"ja":"Proc. First International Workshop on Camera-Based Document Analysis and Recognition (CBDAR2005)"},"pages":"87-94","year":2005,"month":8,"location":"Seoul, South Korea","review":"reviewed","fields":["document-image-analysis","document-image-retrieval"],"abstract":{"ja":"This paper presents a new method of indexing and retrieval\nof planar objects based on feature points and its application\nto document image retrieval using cameras. As\nthe indexing method we propose a method based on local\ncombinations of projective invariants calculated from feature\npoints. As the retrieval method we employ a voting\ntechnique for efficiency and robustness against erasure of\nfeature points. Experimental results on 10,000 images with\n50 queries show that the method is effective (98% accuracy;\nthe remaining query was ranked at the 5th position among\n10,000) and efficient (0.14 second per query)."}},{"key":"Negishi2005_CBDAR","id":"370","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Kazuya NEGISHI","en":""},{"ja":"Masakazu IWAMURA","en":""},{"ja":"Shinichiro OMACHI","en":""},{"ja":"Hirotomo ASO","en":""}],"title":{"ja":"Isolated Character Recognition by Searching Features in Scene Images"},"journal":{"ja":"Proc. First International Workshop on Camera-Based Document Analysis and Recognition (CBDAR2005)"},"pages":"140-147","year":2005,"month":8,"location":"Seoul, South Korea","review":"reviewed","fields":["pattern-recognition","document-image-analysis"],"abstract":{"ja":"Conventional segmentation technique cannot extract an\nisolated character and a touching character. In this paper,\nto utilize information of such characters, we propose a novel\ncharacter recognition method based on extracting feature\npoints and voting. The voting algorithm of the proposed\nmethod is similar to the generalized Hough transform. This\nmethod enables us to extract and recognize such troublesome\ncharacters in relatively shorter computational time.\nThe effectiveness of the proposed method is confirmed by\nexperiments."}},{"key":"Uchida2005_CBDAR","id":"368","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Seiichi Uchida","en":""},{"ja":"Masakazu Iwamura","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Koichi Kise","en":""}],"title":{"ja":"Data Embedding for Camera-Based Character Recognition"},"journal":{"ja":"Proc. First International Workshop on Camera-Based Document Analysis and Recognition (CBDAR2005)"},"pages":"60-67","year":2005,"month":8,"location":"Seoul, South Korea","review":"reviewed","fields":["pattern-recognition","document-image-analysis"],"abstract":{"ja":"In this paper, the embedment of class information into\neach character image is investigated for camera-based\ncharacter recognition as easy and accurate as bar-code\nreading. Each character image is printed with a horizontal\nstripe pattern, called a cross ratio pattern, and the class\ninformation is represented as a cross ratio derived from the\npattern. Since the cross ratio is invariant to projective distortion,\nthe class information is extracted correctly regardless\nof camera angle. Experimental results showed that the\ncross ratios are extracted from distorted character images\nwith very high accuracy and thus very effective to attain\nhigh character recognition rates."}},{"key":"Kise2005b","id":"391","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Nobuhiro Ohmae","en":""}],"title":{"ja":"Mining Tables on the Web for Finding Attributes of a Specified Topic"},"journal":{"ja":"Proceedings of Web Document Analysis 2005"},"pages":"17-20","year":2005,"month":8,"review":"reviewed","fields":["information-retrieval"]},{"key":"Kise2005a","id":"388","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Nobuyuki Kondo","en":""},{"ja":"Tomohiro Nakai","en":""},{"ja":"Keinosuke Matsumoto","en":""}],"title":{"ja":"Analysis of Annotations on Documents for Recycling Information in Collaborative Business Activities"},"journal":{"ja":"Proceedings of the 11th International Conference on Human-Computer Interaction"},"volume":"2","pages":"1-10","year":2005,"month":7,"review":"reviewed","fields":["document-image-analysis","document-image-retrieval"]},{"key":"Takeuchi2005_Conf","id":"374","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"竹内 謹治","en":"Kinji Takeuchi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"}],"title":{"ja":"類似画像を用いた画像の説明文の抽出手法"},"journal":{"ja":"平成17年度電気関係学会関西支部連合大会"},"presen_id":"G3-3","year":2005,"month":11,"review":"not_reviewed","fields":["information-retrieval"],"awards":{"ja":["平成17年度電気関係学会関西支部連合大会奨励賞受賞"]}},{"key":"Sakai2005","id":"372","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"酒井 恵","en":""},{"ja":"江崎弘記","en":""},{"ja":"内田誠一","en":""},{"ja":"岩村雅一","en":""},{"ja":"大町真一郎","en":""},{"ja":"黄瀬浩一","en":""}],"title":{"ja":"カメラによる文字認識のための付加情報抽出法に関する検討"},"journal":{"ja":"電気関係学会九州支部連合大会講演論文集"},"presen_id":"13-2P-03","year":2005,"month":9,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Tanaka2005","id":"381","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"田中 一弘","en":""},{"ja":"内田 誠一","en":""},{"ja":"岩村 雅一","en":""},{"ja":"大町 真一郎","en":""},{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"ペン型デバイスによる手書きパターンへの情報埋め込みと情報抽出 —情報抽出のためのストロークリカバリに関する一検討—"},"journal":{"ja":"電気関係学会九州支部連合大会講演論文集"},"presen_id":"13-2P-11","year":2005,"month":9,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Iwamura2005_MIRU_Isolated","id":"361","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村雅一","en":"Masakazu Iwamura"},{"ja":"根岸和也","en":"Kazuya Negishi"},{"ja":"大町真一郎","en":"Shinichiro Omachi"},{"ja":"阿曽弘具","en":"Hirotomo Aso"}],"title":{"ja":"特徴点探索と投票による孤立文字認識","en":"Isolated Character Recognition by Searching Feature Points and Voting"},"journal":{"ja":"画像の認識・理解シンポジウム2005論文集","en":"Proceedings of MIRU 2005"},"pages":"559-566","year":2005,"month":7,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"通常，情景中の文字情報を利用するためには文字認識が必要である．\nしかし，その前段階である文字切り出しの一般的な手法は\n文字列を成さない「孤立文字」，レイアウトが複雑な数式中の文字，\n隣接した2文字が接触した接触文字等を切り出すことができない．\n従って，これらの文字を認識し，情報を得ることができない．\n\n本稿ではこのような対象からも文字情報が得られるように，\n特徴の検索と投票に基づく，\n文字の切り出しと認識を同時に行う手法を提案する．\n本手法で用いる投票は一般化Hough変換と類似の方法である．\n提案手法は正しく切り出された文字画像に対する個別文\n字認識手法を切り出されていない文字画像にも適用できるように一般化し\nた手法と捉えることができる．\n\n提案手法を実際の画像に適用したところ，\n辞書として用いるフォントと認識対象のフォントが異なる場合でも，\n孤立文字である時計の文字盤や，数式中の接触文字を良好に切り出すことがで\nきた．","en":"To utilize a character information in scene images,\ncharacter recognition and segmentation are required.\nCharacter segmentation is essential preliminary of character recognition.\nHowever, most character segmentation methods cannot extract\n{\\it isolated characters} which do not constitute a character string, \ncharacters in a complex equation,\ntouching characters where two characters are connected and so on.\nTherefore, they cannot be recognized.\n\nIn this paper,\nto utilize information of such characters,\nwe propose a novel recognition method based on extracting feature\npoints and voting. \nThe voting algorithm of the proposed method is similar to the\ngeneralized Hough transform.\nThe proposed method generalizes the conventional \ncharacter recognition for character images segmented correctly \nto one for {\\it unsegmented} character images.\n\nIn experiments, the proposed method extracted\nisolated characters in the face of a clock and touching characters\nwell even if the fonts of a reference image and a target one were\ndifferent."}},{"key":"Iwamura2005_MIRU_Realization","id":"362","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村雅一","en":"Masakazu Iwamura"},{"ja":"内田誠一","en":"Seiichi Uchida"},{"ja":"大町真一郎","en":"Shinichiro Omachi"},{"ja":"黄瀬浩一","en":"Koichi Kise"}],"title":{"ja":"情報付加による認識率100%の実現 —人にも機械にも理解可能な情報伝達のために—","en":"Realization of 100% Recognition Rate with Supplementary Information —For Seamless Man-Machine Communication—"},"journal":{"ja":"画像の認識・理解シンポジウム2005論文集","en":"Proceedings of MIRU 2005"},"pages":"901-908","year":2005,"month":7,"review":"not_reviewed","fields":["pattern-recognition"],"awards":{"ja":["MIRU2005 インタラクティブセッション優秀賞受賞"],"en":["MIRU2005 Best Interactive Session Paper Award Winner"]},"abstract":{"ja":"パターン認識の究極の目標は認識率100%の達成であるが，\n誤認識が全く起こらない識別器の設計は容易ではない．\nそこで，\n本報告では認識誤りを補うために正解クラスに関する情報(付加情報)をパ\nターンとは別に用意し，パターンと付加情報を識別器に同時に与えることで\n認識率100%かつリジェクト率0%を達成する方式を提案する．\nそして，提案する方式で達成可能な認識率やリジェクト率と\nそれに必要な情報量の関係を求める．\n実験の結果，印刷されたかな・漢字3036字種ではオープン実験で4ビット，\nクローズ実験で1ビットの情報の付加で認識率100%かつリジェクト率0%を達\n成できることが示された．\nまた，認識を全く行わずに文字コードの情報をそのまま使用する場合に\n比べると，\n提案する方式で必要な情報量は約1/12，QRコードの面積換算では約1/9である．\n従って，提案する方式は従来のパターン認識のみを用いる場合と\n認識を行わずコードのみの場合を折衷する新しい枠組みといえる．","en":"The ultimate dream in pattern recognition \nis to achieve a 100% recognition rate.\nHowever, it is not so easy.\nIn this report, for achieving a 100% recognition rate with a 0%\nrejection rate, we propose a new framework that the discriminator\nreceives not only a pattern itself but also supplementary information\nthat assists recognition.\nWe clarify the relationships among\na recognition rate, a rejection rate and quantity of supplementary\ninformation.\nFor printed characters, experiments showed that\n4 bits were required in the leave-one-out (L) method\nand 1 bit was in the resubstitution (R) method\nto achieve a 100% recognition rate with a 0% rejection rate.\n\nFurthermore,\ncompared with identification codes such as a character code, \nthe proposed approach compresses about\n1/12 in quantity of supplementary information, and\nabout 1/9 in size of area of QR code.\nThus, the proposed framework combines advantages of\nthe conventional approach of pattern recognition and an identification\ncode."}},{"key":"Nakai2005_MIRU","id":"360","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"複比の投票に基づく文書画像検索と射影歪み補正","en":"Document Image Retrieval and Removal of Perspective Distortion Based on Voting for Cross-Ratios"},"journal":{"ja":"画像の認識・理解シンポジウム2005論文集","en":"Proceedings of MIRU 2005"},"pages":"538-545","year":2005,"month":7,"review":"not_reviewed","fields":["document-image-retrieval","document-image-analysis"],"abstract":{"ja":"本稿では，デジタルカメラで撮影した文書画像の一部あるいは全部を検索質問として，それを含む文書画像をデータベースから検索するという新しい文書画像検索手法を提案する．このような処理の実現には，撮影時に生じる射影歪みに対処しつつ，部分的な手がかりでも柔軟に照合する必要がある．提案手法では，(1)射影変換の不変量である複比を用いたインデックス付け，(2)ハッシュを用いた投票処理の2特徴により，この問題に対処する．また本稿では，文書画像検索の結果として得られる点対応を用いた射影歪みの補正手法についても提案する．提案手法は，我々が別途構築中のデジタルカメラを用いた書き込み抽出システムに用いられるものである．上記２手法について，高解像度デジタルカメラと携帯電話付属デジタルカメラを用いた実験により，有効性の検証を行う．","en":"We propose a method of document image retrieval using digital cameras. The proposed method takes as input parts or the whole of documents acquired as queries by a digital camera, and retrieve document images that include the queries. For this purpose, it is required to solve the problem of \"perspective distortion\" of images, as well as to establish a way for matching parts of document images flexibly. These are achieved based on the following characteristic points of the proposed method: (1) Indexing of document images using the projective invariant called the \"cross-ratio\", (2) Voting with hashes whose indices are defined by the cross-ratios. We also propose a method of removal of perspective distortion using point correspondence obtained through the retrieval. This method is utilized for extracting annotation from document images. From experimental results using digital cameras with high resolution and a camera phone, we demonstrate the effectiveness of the proposed method."}},{"key":"Uchida2005_MIRU","id":"359","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"内田誠一","en":"Seiichi Uchida"},{"ja":"岩村雅一","en":"Masakazu Iwamura"},{"ja":"大町真一郎","en":"Shinichiro Omachi"},{"ja":"黄瀬浩一","en":"Koichi Kise"}],"title":{"ja":"カメラによる文字認識のための付加情報の埋め込みに関する検討","en":"Data Embedding for Camera-Based Character Recognition"},"journal":{"ja":"画像の認識・理解シンポジウム2005論文集","en":"Proceedings of MIRU 2005"},"pages":"235-242","year":2005,"month":7,"review":"not_reviewed","fields":["pattern-recognition","document-image-analysis"],"abstract":{"ja":"デジタルカメラの小型化，高解像度化，また低価格化に伴う普及に伴い，実環境\nでの文字認識・文書画像解析への期待が高まりつつある．本研究では，バーコード\nと同程度の精度で実環境中の文字／文書を認識することを目標として，文字そ\nのものに機械可読性を補強するような情報を埋め込むことを検討している．\n本論文では，撮影の際の射影変換に対応すべく，文字画像に対し，帯状のパターン\nからなる縞模様を埋め込んだを提案する．この隣接する帯の幅から計算される\n複比は射影変換不変，すなわちどのような撮影角度でも常に一定である．従って，\n各文字種について，あらかじめ決まった帯の幅でパターンを埋め込んでおけば，\nそれから計算される複比を識別の手がかりとして認識時に利用できる．シミュ\nレーション\n実験の結果，複比情\n報と文字形状情報を併用することで，射影変換を受けても非常に高い認識精度が\n得られることが判った．","en":"In this paper, we propose a data embedding method for camera-based \ncharacter recognition as easy and accurate as bar-code reading.\nSpecifically, a horizontal stripe pattern, called cross ratio pattern, \nis embedded into a font image of a character category. Since the cross \nratio obtained from the pattern is invariant to projective distortion, \nit always provides correct category candidates regardless of camera angle.\nExperimental results showed that (i) the cross ratio can be extracted from \ndistorted character images with very high accuracy and thus (ii) the \ncross ratio can be used for enhancing recognition performance \nof conventional matching-based recognizers."}},{"key":"Kondo2005","id":"387","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"近藤 伸幸","en":"Nobuyuki Kondo"},{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"松本 啓之亮","en":"Keinosuke Matsumoto"}],"title":{"ja":"デジタルカメラを用いた文書画像からの書き込み抽出手法","en":"A Method for Extracting Annotations from Document Images Captured with Digital Cameras"},"journal":{"ja":"電子情報通信学会2005年総合大会"},"presen_id":"D-12-62","year":2005,"month":3,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"]},{"key":"Nakai2005_IEICE","id":"350","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"松本 啓之亮","en":"Keinosuke Matsumoto"}],"title":{"ja":"デジタルカメラを用いた文書画像検索の一手法","en":"A Method of Document Image Retrieval with Digital Cameras"},"journal":{"ja":"電子情報通信学会2005年総合大会","en":"Proceedings of the 2005 IEICE General Conference"},"presen_id":"D-12-63","year":2005,"month":3,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"]},{"key":"Takeuchi2005_IEICE","id":"351","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"竹内 謹治","en":"Kinji Takeuchi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"松本 啓之亮","en":"Keinosuke Matsumoto"}],"title":{"ja":"語の共起の統計情報を用いた画像と説明文の対応付け","en":"Linking Images with Their Expository Text Based on Statistical Information of Word Co-occurrence"},"journal":{"ja":"電子情報通信学会2005年総合大会","en":"Proceedings of the 2005 IEICE General Conference"},"presen_id":"D-5-1","year":2005,"month":3,"review":"not_reviewed","fields":["information-retrieval"]},{"key":"Shime2005","id":"386","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"注連 隆夫","en":"Takao Shime"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"松本 啓之亮","en":"Keinosuke Matsumoto"}],"title":{"ja":"画像検索のための視覚化インタラクティブ・モジュール","en":"Interactive Module for Visualizing Results of Image Retrieval"},"journal":{"ja":"電子情報通信学会2005年総合大会講演論文集"},"presen_id":"D-5-2","year":2005,"month":3,"review":"not_reviewed","fields":["information-retrieval"]},{"key":"Iwamura2005_IEICEPRMU","id":"346","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu IWAMURA"},{"ja":"内田 誠一","en":"Seiichi UCHIDA"},{"ja":"大町 真一郎","en":"Shinichiro OMACHI"},{"ja":"黄瀬 浩一","en":"Koichi KISE"}],"title":{"ja":"認識が与える情報量—認識率100%まであと何ビット？—","en":"Quantity of Information of Recognition —How Many Bits Are Lacking for 100% Recognition?—"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"104","number":"PRMU-742","presen_id":"PRMU2004-237","pages":"35-40","year":2005,"month":3,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"パターン認識の究極の目標は認識率100%の達成であるが，誤認識が全く起こらない識別器の設計は容易ではない．本報告では認識誤りを補うために正解クラスに関する情報(付加情報)をパターンとは別に用意し，パターンと付加情報を識別器に同時に与えることで認識率100%かつリジェクト率0%を達成する方式を提案する．実験の結果，印刷文字ではオープン実験で4ビット，クローズ実験で1ビットの情報の付加で達成できることが示された．このような情報量は認識率や認識後に残る不確かさとは別の性質を持ち，識別器の新たな評価基準となり得る．","en":"The ultimate dream in pattern recognition is to achieve 100% of recognition rate. However, it is not so easy. In this report, for achieving 100% of recognition rate and 0% of rejection rate, we propose a new framework that the discriminator receives not only a pattern itself but also supplementary information about the class that the pattern belongs to. For printed characters, experiments showed that 4 bits are required in the leave-one-out (L) method and 1 bit is in the resubstitution (R) method. Such kind of quantity of information has different characteristics from recognition rates and ambiguity after recognition. This criterion can be a new criterion of a discriminator."}},{"key":"Kise2005_IEICEPRMU","id":"348","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":"Koichi KISE"},{"ja":"大町 真一郎","en":"Shinichiro OMACHI"},{"ja":"内田 誠一","en":"Seiichi UCHIDA"},{"ja":"岩村 雅一","en":"Masakazu IWAMURA"}],"title":{"ja":"カメラを用いた文字認識・文書画像解析の現状と課題","en":"Current Status and Future Prospects of Camera-Based Character Recognition and Document Image Analysis"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"104","number":"PRMU-742","presen_id":"PRMU2004-246","pages":"85-90","year":2005,"month":3,"review":"not_reviewed","fields":["pattern-recognition","document-image-analysis"],"abstract":{"ja":"デジタルカメラの小型化，高解像度化，一般化に伴い，これを入力手段として用いた文字認識・文書画像解析への期待が高まりつつある．本稿では，デジタルカメラを用いた文字認識・文書画像解析について現在の代表的な技術を俯瞰するとともに，バーコードなどの関連技術との比較によって，文字認識，文書画像解析の立場や将来への課題を浮き彫りにする．また，将来に向けた試みの一つとして，著者らが「バーコードリーダ並みに手軽で高精度な文字認識」を目指して行っている研究「複比を用いた文字への情報埋め込み」の一端についても紹介する．","en":"Pervasive use of handy digital cameras with higher resolution is now defining new roles of character recognition and document image analysis\nas a mean of analyzing camera-captured images. In this report, we survey state-of-the-art of research and technologies of camera based character recognition and document image analysis. We also describe the current position and future prospects of character recognition and document image analysis in comparison with related technologies such as ``barcodes''. In addition, we briefly introduce our research entitled\n``embedding information on characters using cross ratios'' whose final goal is to make character recognition as easy and accurate as bar-code reading."}},{"key":"Nakai2005_IEICEPRMU","id":"349","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"中居 友弘","en":"Tomohiro Nakai"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"松本 啓之亮","en":"Keinosuke Matsumoto"}],"title":{"ja":"複比とハッシュに基づく文書画像検索","en":"Document Image Retrieval Based on Cross-Ratio and Hashing"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"104","number":"PRMU-742","presen_id":"PRMU2004-249","pages":"103-108","year":2005,"month":3,"review":"not_reviewed","fields":["document-image-analysis","document-image-retrieval"],"abstract":{"ja":"本稿では，デジタルカメラを用いた新しい文書画像検索手法を提案する．提案手法は，デジタルカメラで撮影した文書画像の一部あるいは全部を検索質問として，それを含む文書画像をデータベースから検索するものである．これを実現するためには，撮影時に生じる射影変換歪みに対処しつつ，部分的な手がかりでも柔軟に照合する必要がある．この問題に対して，提案手法では，(1)射影変換の不変量である複比を用いたインデックス付け，(2)ハッシュを用いた投票処理の２特徴により対処する．高解像度デジタルカメラと携帯電話付属のデジタルカメラで撮影した画像を用いた実験により，提案手法の有効性を検証する．","en":"In this report, we propose a new method of document image retrieval using digital cameras. The proposed method is to retrieve from document images which include parts or the whole of document images acquired by a digital camera. For this purpose, it is required to solve the problem of ``perspective distortion'' of images, as well as to establish a way for matching parts of document images flexibly. These are achieved based on the following characteristic points of the proposed method: (1) Indexing of document images using the projective invariant called the ``cross-ratio'', (2) Voting with hashes whose indices are defined by the cross-ratios. From experimental results using digital cameras with high resolution and a camera phone, we demonstrate the effectiveness of the proposed method."}},{"key":"Negishi2005_IEICEPRMU","id":"347","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"根岸 和也","en":"Kazuya Negishi"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"阿曽 弘具","en":"Hirotomo Aso"}],"title":{"ja":"特徴点探索による孤立文字認識","en":"Isolated Character Recognition by Searching Feature Points"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"104","number":"PRMU-742","presen_id":"PRMU2004-238","pages":"41-46","year":2005,"month":3,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"画像に含まれている文字の情報は，非常に利用価値の高いものであるが，その文字が時計盤のように孤立している場合は，従来手法では切り出すことができないため，有効に利用することができない．本稿では，このような孤立文字に対して，文字の持つ本質的な特徴を利用して，切り出しと認識を同時に行う手法を提案する．具体的には，特徴点として角と曲線部分を検出し，使用する．特徴点の探索には，文字のエッジの方向を利用した．これらの特徴点の探索結果を総合的に判断することで文字の切り出し箇所を決定する．また，従来法では参照する画像と認識したい文字のフォントが異なる場合は，認識率が低下してしまう問題が存在したが，実験により提案手法がこの問題に有効であることを確認した．","en":"A character information in an image is valuable.However, if a character is isolated as in the face of a clock, it is not easy to extract the character region.We propose a method to identify an isolated character by extracting feature points of characters. We use corners and curves as the feature points.Using edge direction of each point of an image, these feature pointsare extracted, and the position of a character is determined.The effectiveness of the proposed method is shown by by experiments."}},{"key":"Iwamura2005","id":"329","type":"other","language":"EN","show":"show_except_datafile","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Hirotomo Aso","en":""}],"title":{"ja":"On the Bias of Predictive Distribution in Pattern Recognition"},"journal":{"ja":"Systems and Computers in Japan"},"volume":"36","number":"5","pages":"45-54","year":2005,"month":5,"publisher":"John Wiley & Sons, Inc.","review":"not_reviewed","fields":["pattern-recognition"],"note":{"ja":"PDF file is available at <a href=\"http://www3.interscience.wiley.com/cgi-bin/jissue/110429145\">http://www3.interscience.wiley.com/cgi-bin/jissue/110429145</a>."},"abstract":{"ja":"To estimate distribution from training samples,\nthe maximum likelihood estimation treats the unknown parameter of the\ndistribution as a constant. On the other hand, Bayesian estimation\ntreats the parameter as a random variable.\nIn pattern recognition, Bayesian estimation has been known to improve\nrecognition accuracy.\nHowever, it was pointed out that the Bayesian estimation is not\neffective due to the bias of the likelihood when sample sizes of\nclasses are not the same.\nIn this paper, we show recognition accuracy is improved by modifying\nthe bias of the likelihood when sample sizes are not the same.\nThis indicates that the cause of the ineffectiveness is the bias.\nWe derive the formula of the bias of\nGeisser's predictive distribution without any approximation, and show\na way of modification of the bias of the likelihood.\nWe confirm effectiveness of the proposed method in experiments.\nIn addition, the derived formula gives the theoretical background of a\nknown empirical knowledge."}},{"key":"Yoshitani2004_Trans","id":"414","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"吉谷 仁志","en":"Hitoshi Yoshitani"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"松本 啓之亮","en":"Keinosuke Matsumoto"}],"title":{"ja":"サポートベクトルマシンを用いた新聞記事からのプロフィール情報抽出","en":"Extraction of Profile Information from Newspaper Articles Using Support Vector Machines"},"journal":{"ja":"電気学会論文誌（Ｃ）","en":"Transactions of the Institute of  Electrical Engineers of Japan"},"volume":"124","number":"11","pages":"2260-2266","year":2004,"month":11,"review":"reviewed","fields":[]},{"key":"Smith2004","id":"412","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"E. B. Smith","en":""},{"ja":"D. Monn","en":""},{"ja":"H. Veeramachaneni","en":""},{"ja":"K. Kise","en":""},{"ja":"A. Malizia","en":""},{"ja":"L. Todoran","en":""},{"ja":"A. El-Nasan","en":""},{"ja":"R. Ingold","en":""}],"title":{"ja":"Reports of the DAS02 Working Groups"},"journal":{"ja":"International Journal of Document Analysis and Recognition"},"volume":"6","number":"3","pages":"211-217","year":2004,"month":3,"publisher":"Springer","review":"reviewed","fields":[]},{"key":"Iwamura2004_IEICE","id":"304","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"阿曽 弘具","en":"Hirotomo Aso"}],"title":{"ja":"パターン認識における予測分布の偏りに関する一考察","en":"On the Bias of Predictive Distribution in Pattern Recognition"},"journal":{"ja":"電子情報通信学会論文誌D-II","en":"Trans. IEICE"},"volume":"J87-D-II","number":"2","pages":"407-415","year":2004,"month":2,"review":"reviewed","fields":["pattern-recognition"],"url":"http://search.ieice.org/bin/summary.php?id=j87-d2_2_407","note":{"ja":"論文のPDFファイルを<a href=\"http://www.ieice.org/jpn/trans_online/\">電子情報通信学会 Transactions online</a>からダウンロードすることができます．\n<br>\n<a href=\"http://www.m.cs.osakafu-u.ac.jp/cgi-bin/publication_public.cgi?func=show_detail&id=329&prev_func=article_list&lang=en\">英語の翻訳</a>があります。","en":"PDF file is available at <a href=\"http://www.ieice.org/jpn/trans_online/\">IEICE Transactions online (in Japanese)</a>.\n<br>\n<a href=\"http://www.m.cs.osakafu-u.ac.jp/cgi-bin/publication_public.cgi?func=show_detail&id=329&prev_func=article_list&lang=en\">The paper is translated into English</a>."},"abstract":{"ja":"学習用サンプルから分布を推定する際，\n未知パラメータを定数として推定する最尤推定に対し，\n未知パラメータを確率変数とするベイズ推定がある．\nベイズ推定を用いたパターン認識は，クラス間でサンプル数が等しい場合\nは有効であるが，等しくない場合には尤度の偏りのために\n認識性能が向上しないことが指摘されている．\nクラス間でサンプル数が異なる場合のパターン認識では\n尤度の偏りに差が生じ，この偏りの差が認識性能を悪化させていると考えられ\nる．\n本論文では，尤度の偏りを補正することで認識性能が改善できると考え，\n予測分布の偏りを理論的に導き，厳密解を与える．\n予測分布の偏りの理論式が得られたことにより，尤度の補正が可能となる．\n実際に尤度の補正法を提案し，これにより\n認識性能が改善することを検証する．\nさらに本論文で導く理論式から，\nこれまで経験的に得られた知見の理論的根拠を明らかにする．","en":"Bayesian estimation is known to improve recognition accuracy.\nHowever, the method is not effective due to the bias of\nthe probability when sample sizes are not the same.\nIn this paper, the function of the bias of Geisser's predictive\ndistribution is derived without any approximation.\nIt is shown that correcting likelihood by the function improves\nrecognition accuracy.\nThis indicates the bias is one of the cause of decreasing recognition\naccuracy.\nFrom the character of the function, it is easy to explain the\nknowledges obtained from experiments.\nThe derivation of the function can be applied to other predictive\ndistributions."}},{"key":"Kise2004b","id":"415","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Shota Fukushima","en":""},{"ja":"Keinosuke Matsumoto","en":""}],"title":{"ja":"Document Image Retrieval in a Question Answering System for Document Images"},"journal":{"ja":"6th International Workshop on Document Analysis Systems"},"book_title":"Document Analysis Systems VI","volume":"3163","pages":"521-532","year":2004,"month":9,"publisher":"Springer","review":"reviewed","fields":[]},{"key":"Raudys2004_SSPR","id":"330","type":"intl_conf","language":"JA","show":"show","authors":[{"ja":"Sarunas Raudys","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Structures of Covariance Matrix in Handwritten Character Recognition"},"journal":{"ja":"Lecture Notes in Computer Science (Joint IAPR International Workshops SSPR 2004 and SPR 2004)"},"volume":"3138","pages":"725-733","year":2004,"month":8,"location":"Lisbon, Portugal","review":"reviewed","fields":["pattern-recognition"],"doi":"10.1007/978-3-540-27868-9_79","abstract":{"ja":"The integrated approach is a classifier established on statistical estimator and artificial neural network. This consists of preliminary data whitening transformation which provides good starting weight vector, and fast training of single layer perceptron (SLP). If sample size is extremely small in comparison with dimensionality, this approach could be ineffective. In the present paper, we consider joint utilization of structures and conventional regularization techniques of sample covariance matrices in order to improve recognition performance in very difficult case where dimensionality and sample size do not differ essentially. The techniques considered reduce a number of parameters estimated from training set. We applied our methodology to handwritten Japanese character recognition and found that combination of the integrated approach, conventional regularization and various structurization methods of covariance matrix outperform other methods including optimized Regularized Discriminant Analysis (RDA)."}},{"key":"Raudys2004_MCS","id":"312","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Sarunas Raudys","en":""},{"ja":"Masakazu Iwamura","en":""}],"title":{"ja":"Multiple Classifiers System for Reducing Influences of Atypical Observations"},"journal":{"ja":"Fifth International Workshop on Multiple Classifier Systems (MCS 2004)"},"volume":"3077","pages":"233-242","year":2004,"month":6,"location":"Cagliari, Italy","review":"reviewed","fields":["pattern-recognition"],"doi":"10.1007/978-3-540-25966-4_23","abstract":{"ja":"Atypical observations, which are called outliers, are one of difficulties to apply standard Gaussian density based pattern classification methods. Large number of outliers makes distribution densities of input features multimodal. The problem becomes especially challenging in high-dimensional feature space. To tackle atypical observations, we propose multiple classifiers systems (MCSs) whose base classifiers have different representations of the original feature by transformations. This enables to deal with outliers in different ways. As the base classifier, we employ the integrated approach of statistical and neural networks. This consists of data whitening and training of single layer perceptron (SLP). Data whitening makes marginal distributions close to unimodal, and SLP is robust to outliers. Various kinds of combination strategies of the base classifiers achieved reduction of generalization error in comparison with the benchmark method, the regularized discriminant analysis (RDA)."}},{"key":"Negishi2004_FIT","id":"339","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"根岸 和也","en":"Kazuya Negishi"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"阿曽 弘具","en":"Hirotomo Aso"}],"title":{"ja":"部分領域の探索による孤立文字認識","en":"Isolated Character Recognition by Search of Partial Regions"},"journal":{"ja":"第3回情報科学技術フォーラム(FIT2004)","en":"FIT2004"},"presen_id":"I-018","year":2004,"month":10,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Takeya2004_FIT","id":"340","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"竹谷 仁志","en":"Hitoshi Takeya"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"下村 正夫","en":"Masao Shimomura"},{"ja":"阿曽 弘具","en":"Hirotomo Aso"}],"title":{"ja":"単純再帰ネットワークの定量的な評価方法の検討","en":"Quantitative Evaluation of Simple Recurrent Network"},"journal":{"ja":"第3回情報科学技術フォーラム(FIT2004)","en":"FIT2004"},"presen_id":"H-005","year":2004,"month":9,"review":"not_reviewed","fields":[]},{"key":"Mochida2004","id":"343","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"持田 広志","en":"Hiroshi Mochida"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"阿曽 弘具","en":"Hirotomo Aso"}],"title":{"ja":"特徴点の追跡による二次元画像列からの三次元形状復元"},"journal":{"ja":"平成１６年度 電気関係学会東北支部連合大会"},"presen_id":"2H-19","year":2004,"month":8,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Negishi2004_Conf","id":"341","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"根岸 和也","en":"Kazuya Negishi"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"阿曽 弘具","en":"Hirotomo Aso"}],"title":{"ja":"切り出しを同時に行う孤立文字認識法"},"journal":{"ja":"平成16年度電気関係学会東北支部連合大会"},"presen_id":"2H-13","year":2004,"month":8,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Takeya2004_Conf","id":"342","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"竹谷 仁志","en":"Hitoshi Takeya"},{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"下村 正夫","en":"Masao Shimomura"},{"ja":"阿曽 弘具","en":"Hirotomo Aso"}],"title":{"ja":"単純再帰ネットワークを用いた言語獲得過程の評価"},"journal":{"ja":"平成16年度電気関係学会東北支部連合大会"},"presen_id":"0.000001","year":2004,"month":8,"review":"not_reviewed","fields":[]},{"key":"Ohmae2004","id":"380","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"大前 信弘","en":""},{"ja":"吉谷 仁志","en":""},{"ja":"黄瀬 浩一","en":""},{"ja":"松本 啓之亮","en":""}],"title":{"ja":"Webを対象としたプロフィール情報抽出の基礎的検討"},"journal":{"ja":"情報処理学会第66回全国大会"},"year":2004,"month":3,"review":"not_reviewed","fields":[]},{"key":"Fukushima2004","id":"417","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"福嶋 祥太","en":"Shota Fukushima"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"松本 啓之亮","en":"Keinosuke Matsumoto"}],"title":{"ja":"文書画像に対する質問応答システムIQAS"},"journal":{"ja":"電子情報通信学会技術研究報告"},"presen_id":"PRMU2003-217","pages":"13-18","year":2004,"month":2,"review":"not_reviewed","fields":[]},{"key":"Yoshitani2004_SIG","id":"418","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"吉谷 仁志","en":"Hitoshi Yoshitani"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"松本 啓之亮","en":"Keinosuke Matsumoto"}],"title":{"ja":"サポートベクトルマシンを用いたプロフィール情報の抽出"},"journal":{"ja":"情報処理学会研究報告"},"presen_id":"2004-NL-159","pages":"7-12","year":2004,"month":1,"review":"not_reviewed","fields":[]},{"key":"Kise2004_Symp","id":"416","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":""},{"ja":"福嶋 祥太","en":""},{"ja":"松本 啓之亮","en":""}],"title":{"ja":"文書画像検索における精度向上の一手法","en":"A Method of Improving Precision for Document Image Retrieval"},"journal":{"ja":"画像の認識・理解シンポジウム講演論文集"},"volume":"2","pages":"483-488","year":2004,"review":"not_reviewed","fields":[]},{"key":"Kise2004a","id":"411","type":"book","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Markus Junker","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"Keinosuke Matsumoto","en":""}],"title":{"ja":"Passage Retrieval Based on Density Distributions of Terms and Its Applications to Document Retrieval and Question Answering"},"book_title":"Reading and Learning","pages":"306-327","year":2004,"month":4,"publisher":"Springer","review":"not_reviewed","fields":[]},{"key":"Babaguchi2004","id":"410","type":"book","language":"JA","show":"show","authors":[{"ja":"馬場口 登","en":""},{"ja":"黄瀬 浩一","en":""},{"ja":"山岡 正輝","en":""}],"title":{"ja":"第5章「文書画像解析」"},"book_title":"画像解析ハンドブック， 機能編?部応用","pages":"1813-1833","year":2004,"publisher":"東京大学出版会","review":"not_reviewed","fields":[]},{"key":"Iwamura2004","id":"303","type":"other","language":"EN","show":"show_except_datafile","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Hirotomo Aso","en":""}],"title":{"ja":"Estimation of True Mahalanobis Distance from Eigenvectors of Sample Covariance Matrix"},"journal":{"ja":"Systems and Computers in Japan"},"volume":"35","number":"9","pages":"30-38","year":2004,"month":8,"publisher":"John Wiley & Sons, Inc.","review":"not_reviewed","fields":["pattern-recognition"],"note":{"ja":"PDF file is available at <a href=\"http://www3.interscience.wiley.com/cgi-bin/jissue/109076289\">http://www3.interscience.wiley.com/cgi-bin/jissue/109076289</a>."},"abstract":{"ja":"In statistical pattern recognition,\nthe Bayesian decision theory gives a decision to minimize the expected\nprobability of misclassification as long as the true distributions\nare given.\nHowever, in most practical situations, the true distributions are\nunknown, and the parameters of the distributions are usually estimated\nfrom training sample vectors.\nIt is well-known that estimated parameters contain estimation errors\nwhen sample size is small,\nand the errors cause bad influence on recognition performance.\nAmong the estimation errors of parameters,\nthe estimation errors of eigenvectors have not been considered enough.\nIn this paper, we present a method to estimate the true Mahalanobis\ndistance from the sample eigenvectors\n(the eigenvectors of sample covariance matrix)\nby considering the estimation errors of eigenvectors.\nRecognition experiments show that\nthe true Mahalanobis distance can be estimated,\nand better recognition accuracy is achieved\nby applying the proposed method\nwithout many training samples and any hyper-parameters."}},{"key":"Kise2003_IEICE","id":"421","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Markus Junker","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"Keinosuke Matsumoto","en":""}],"title":{"ja":"Effectiveness of Passage-Based Document Retrieval for Short Queries"},"journal":{"ja":"IEICE Trans. on Information and Systems"},"volume":"E86-D","number":"9","pages":"1753-1761","year":2003,"month":9,"review":"reviewed","fields":[]},{"key":"Iwamura2003_IEICE","id":"6","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"阿曽 弘具","en":"Hirotomo Aso"}],"title":{"ja":"標本共分散行列の固有ベクトルを用いた真のマハラノビス距離の推定法","en":"Estimation of True Mahalanobis Distance from Eigenvectors of Sample Covariance Matrix"},"journal":{"ja":"電子情報通信学会論文誌D-II","en":"Trans. IEICE"},"volume":"J86-D-II","number":"1","pages":"22-31","year":2003,"month":1,"review":"reviewed","fields":["pattern-recognition"],"url":"https://search.ieice.org/bin/summary.php?id=j86-d2_1_22","note":{"ja":"論文のPDFファイルを<a href=\"http://www.ieice.org/jpn/trans_online/\">電子情報通信学会 Transactions online</a>からダウンロードすることができます．<br>\n<a href=\"http://www.m.cs.osakafu-u.ac.jp/cgi-bin/publication_public.cgi?func=show_detail&id=303&prev_func=article_list&lang=en\">英語の翻訳</a>があります。","en":"PDF file is available at <a href=\"http://www.ieice.org/jpn/trans_online/\">IEICE Transactions online (in Japanese)</a>.\n<br>\n<a href=\"http://www.m.cs.osakafu-u.ac.jp/cgi-bin/publication_public.cgi?func=show_detail&id=303&prev_func=article_list&lang=en\">The paper is translated into English</a>."},"abstract":{"ja":"パターン認識で用いられる，ベイズの定理から導かれる統計的識別関数は，\nパターンの確率分布が正しく与えられたときに誤識別率が最小になる．\nパターンの確率分布は一般に未知であるため，正規分布を仮定して，\nその平均ベクトルと共分散行列は学習サンプルからの推定値を\n用いることが多い．\nしかし，推定に用いる学習サンプルが不足すると，\n推定値が誤差を含み，認識性能が低下することが知られている．\n平均ベクトルや共分散行列の固有値の推定誤差についてはよく調査され，\n認識性能の低下を回避する方法が提案されている．\nしかし，共分散行列の固有ベクトルの推定誤差については，\nこれまでほとんど考慮されてこなかった．\n本論文では，学習サンプルが十分用意できない場合においても高精度な認識を\n行うことを目的とし，\n共分散行列の固有ベクトルに誤差がある場合でも，\n２次識別関数の主要部分であるマハラノビス距離を\n正しく推定する手法を提案する．\n提案手法を文字認識に適用した結果，パラメータの推定に用いる学習サンプルが\n少ない場合でも，\nマハラノビス距離を正しく推定することができ，\n認識性能の改善が見られた．","en":"In statistical pattern recognition,\nparameters of distributions are\nusually estimated from training sample vectors.\nHowever, estimated parameters contain estimation errors and\nthe errors cause bad influence on recognition\nperformance when the sample size is not sufficient.\nIn this paper, we present a method to estimate the true Mahalanobis\ndistance from sample eigenvectors\n(eigenvectors of sample covariance matrix)\nby considering the errors of eigenvectors.\nRecognition experiments show that by applying the proposed method,\nthe true Mahalanobis distance can be estimated\neven if the sample size is small,\nand better recognition accuracy is achieved."}},{"key":"Kise2003_ITL_Indexing","id":"419","type":"letter","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"松本 啓之亮","en":"Keinosuke Matsumoto"}],"title":{"ja":"物理構造に基づく文書画像の索引付けと検索","en":"Indexing and Retrieval of Document Images Based on Physical Structures"},"journal":{"ja":"情報科学技術レターズ","en":"Information Technology Letters"},"pages":"203-204","year":2003,"month":9,"review":"reviewed","fields":[]},{"key":"Kise2003_ITL_Stippling","id":"420","type":"letter","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"松本 啓之亮","en":"Keinosuke Matsumoto"}],"title":{"ja":"文書余白領域へのデータの点描","en":"Stippling Data on Margins of Documents"},"journal":{"ja":"情報科学技術レターズ","en":"Information Technology Letters"},"pages":"201-202","year":2003,"month":9,"review":"reviewed","fields":[]},{"key":"Kise2003_ICDARa","id":"422","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Yasuo Miki","en":""},{"ja":"Keinosuke Matsumoto","en":""}],"title":{"ja":"Stippling Data on Backgrounds of Pages —Toward Seamless Integration of Paper and Electronic Documents"},"journal":{"ja":"Proc. of the 7th International Conference on Document Analysis and Recognition (ICDAR2003)"},"pages":"1213-1217","year":2003,"month":8,"review":"reviewed","fields":[]},{"key":"Kise2003_ICDARb","id":"423","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Wuotang Yin","en":""},{"ja":"Keinosuke Matsumoto","en":""}],"title":{"ja":"Document Image Retrieval Based on 2D Density Distributions of Terms with Pseudo Relevance Feedback"},"journal":{"ja":"Proc. of the 7th International Conference on Document Analysis and Recognition (ICDAR2003)"},"pages":"488-492","year":2003,"month":8,"review":"reviewed","fields":[]},{"key":"Kise2003","id":"424","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Keinosuke Matsumoto","en":""}],"title":{"ja":"Indexing and Retrieval of Document Images Using Term Positions and Physical Structures"},"journal":{"ja":"Proc. of the Third International Workshop on Document Layout Interpretation and Its Applications"},"pages":"19-22","year":2003,"month":8,"review":"reviewed","fields":[]},{"key":"Iwamura2003_FIT","id":"302","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"阿曽 弘具","en":"Hirotomo Aso"}],"title":{"ja":"部分空間法における原点の位置の影響","en":"The Effect of Positioning the Origin for the Subspace Method"},"journal":{"ja":"第２回情報科学技術フォーラム(FIT2003)","en":"Forum on Information Technology (FIT2003)"},"presen_id":"I-065","year":2003,"month":9,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Iwamura2003_IEICEPRMU","id":"42","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shinichiro Omachi"},{"ja":"阿曽 弘具","en":"Hirotomo Aso"}],"title":{"ja":"次元の縮退を利用した頑健な分布の推定法","en":"Robust Estimation of Distribution by Shrinkage Technique"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"102","number":"652","presen_id":"PRMU2002-216","pages":"31-36","year":2003,"month":2,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"パターン認識では共分散行列の固有値と固有ベクトルを必要とすることが多\nいが，共分散行列を少数の学習用のサンプルから推定する場合には\n固有値の偏りによって認識性能が低下することが知られている．\nこれに対して，小さな固有値が認識性能に悪影響を及ぼすとして\n認識に用いない手法や\n固有値展開後に固有値を補正する手法が提案されている．\nこれらはいずれも固有値展開の後，偏った固有値に対して対処する手法である\nが，固有値の偏りが固有値展開の際に発生することから，\n固有値に誤差が生じにくい工夫を固有値展開の前に施すことも\n可能であると考えられる．\n\n本論文では，最初に固有値の偏りが固有値展開で生じることを確認する．\nそして，固有値展開前に共分散行列を縮退させておくことで，\n固有値展開によって固有値が偏りにくくする手法を提案する．\n提案手法が標本共分散行列を用いた場合に比べて真の分布をより正しく推定し，\n認識性能を改善することを認識実験により確認した．","en":"Most pattern recognition applications require\nthe eigenvalues and eigenvectors of the covariance matrix.\nIt is well known that when the number of training samples is small, \nthe eigenvalues of the covariance matrix contains bias,\nand the bias degrades recognition performance.\nThere are some methods which ignore the small eigenvalues,\nor acquire better estimates of the covariance matrix by\ncorrecting the eigenvalues.\nThough all of these methods cope with the eigenvalues obtained after\neigen decomposition,\neigen decomposition seems to cause the biases of the eigenvalues.\nTherefore, it is worth trying to devise a method which avoids\nbias of eigenvalues.\n\nIn this paper, it is confirmed that biases of the eigenvalues appear\nafter eigen decomposition by experiments.\nThen a method of shrinking the covariance matrix before eigen\ndecomposition for avoiding bias of the eigenvalues are proposed.\nThe ability of the proposed method of estimating the true distribution\nmore precisely than using the sample covariance matrix\n and of improving recognition performance is confirmed by the\n recognition experiments."}},{"key":"Iwamura2003_PhD","id":"164","type":"doctoral_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"パターン認識における確率分布推定法に関する研究","en":"A Study on Estimation of Probability distribution in Pattern Recognition"},"year":2003,"month":3,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Kise2002b","id":"951","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"K. Kise","en":""},{"ja":"M. Tsujino","en":""},{"ja":"K. Matsumoto","en":""}],"title":{"ja":"Spotting Where to Read on Pages — Retrieval of Relevant Parts from Page Images"},"journal":{"ja":"Document Analysis Systems V, Lecture Notes in Computer Science"},"volume":"2423","pages":"388-399","year":2002,"month":8,"review":"reviewed","fields":[]},{"key":"Kise2002a","id":"425","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Masaaki Tsujino","en":""},{"ja":"Keinosuke Matsumoto","en":""}],"title":{"ja":"Spotting Where to Read on Pages - Retrieval of Relevant Parts  from Page Images"},"journal":{"ja":"5th International Workshop on Document Analysis Systems"},"book_title":"Document Analysis Systems V","volume":"2423","pages":"388-399","year":2002,"month":8,"review":"reviewed","fields":[]},{"key":"Iwamura2002","id":"24","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Shinichiro Omachi","en":""},{"ja":"Hirotomo Aso","en":""}],"title":{"ja":"A Method to Estimate the True Mahalanobis Distance from Eigenvectors of Sample Covariance Matrix"},"journal":{"ja":"Lecture Notes in Computer Science (Joint IAPR International Workshops SSPR 2002 and SPR 2002)"},"volume":"2396","pages":"498-507","year":2002,"month":8,"location":"Windsor, Ontario, Canada","review":"reviewed","fields":["pattern-recognition"],"doi":"10.1007/3-540-70659-3_52","abstract":{"ja":"In statistical pattern recognition,\nthe parameters of distributions are\nusually estimated from training sample vectors.\nHowever, estimated parameters contain estimation errors,\nand the errors cause bad influence on recognition\nperformance when the sample size is not sufficient.\nSome methods can obtain better estimates of the eigenvalues\nof the true covariance matrix\nand can avoid bad influences caused by estimation errors of\neigenvalues.\nHowever, estimation errors of eigenvectors of covariance matrix have\nnot been considered enough.\nIn this paper, we consider estimation errors of eigenvectors and show\nthe errors can be regarded as estimation errors of eigenvalues.\nThen, we present a method to estimate the true Mahalanobis\ndistance from eigenvectors of the sample covariance matrix.\nRecognition experiments show that by applying the proposed method,\nthe true Mahalanobis distance can be estimated\neven if the sample size is small,\nand better recognition accuracy is achieved.\nThe proposed method is useful for the practical applications of\npattern recognition since the proposed method is effective\nwithout any hyper-parameters."}},{"key":"Kise2002_PRMU","id":"426","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":""},{"ja":"辻野 雅章","en":""},{"ja":"松本 啓之亮","en":""}],"title":{"ja":"文字の２次元出現密度分布に基づく文書画像の部分検索"},"journal":{"ja":"電子情報通信学会技術研究報告"},"presen_id":"PRMU2002-16","year":2002,"month":5,"review":"not_reviewed","fields":[]},{"key":"Matsumoto2002","id":"947","type":"book","language":"JA","show":"show","authors":[{"ja":"松本啓之亮","en":""},{"ja":"黄瀬浩一","en":""},{"ja":"森直樹","en":""}],"title":{},"book_title":"知能システム工学入門","pages":"34-92","year":2002,"month":8,"publisher":"コロナ社","review":"not_reviewed","fields":[]},{"key":"Kise2002c","id":"427","type":"other","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":""},{"ja":"松本 啓之亮","en":""}],"title":{"ja":"文書画像データベースの検索に関する一考察"},"journal":{"ja":"科学技術分野における電子的情報処理に関する研究集会"},"pages":"91-96","year":2002,"month":12,"review":"not_reviewed","fields":[]},{"key":"Kise2001b","id":"918","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Markus Junker","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"Keinosuke Matsumoto","en":""}],"title":{"ja":"Passage-Based Document Retrieval as a Tool for Text Mining with User's Information Needs"},"journal":{"ja":"Discovery Science,　Lecture Notes in Computer Science"},"volume":"2226","pages":"155-169","year":2001,"month":11,"review":"reviewed","fields":[]},{"key":"Kise2001a","id":"428","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Markus Junker","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"Keinosuke Matsumoto","en":""}],"title":{"ja":"Passage-Based Document Retrieval as a Tool  for Text Mining with User's Information Needs"},"journal":{"ja":"Proc. of 4th International Conference on Discovery Science"},"pages":"155-169","year":2001,"month":11,"review":"reviewed","fields":[]},{"key":"Kise2001_ICDAR","id":"429","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Markus Junker","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"Keinosuke Matsumoto","en":""}],"title":{"ja":"Experimental Evaluation of Passage-Based Document Retrieval"},"journal":{"ja":"Proc. of the 6th International Conference on Document Analysis and Recognition (ICDAR2001)"},"pages":"592-596","year":2001,"month":9,"review":"reviewed","fields":[]},{"key":"Kise2001_Trans","id":"430","type":"survey","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"情報のリサイクル − 情報検索の夢"},"journal":{"ja":"電気学会論文誌(C)"},"volume":"121-C","number":"1","pages":"50-51","year":2001,"month":1,"review":"not_reviewed","fields":[],"note":{"ja":"21世紀の扉を開く記念部門誌，特集解説"}},{"key":"Mizuno2000","id":"916","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"水野 浩之","en":"Hiroyuki Mizuno"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"松本 啓之亮","en":"Keinosuke Matsumoto"}],"title":{"ja":"窓関数を用いた部分テキスト検索 — 固定長法と出現密度法の比較 —","en":"Passage Retrieval Methods Using a Window Function — A Comparison Between the Fixed Length Method and the Density Distribution Method —"},"journal":{"ja":"電気学会論文誌（Ｃ）","en":"Trans. IEE of Japan"},"volume":"120-C","number":"11","pages":"1539-1546","year":2000,"month":11,"review":"reviewed","fields":[]},{"key":"Iwamura2000_ICPR","id":"29","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Masakazu Iwamura","en":""},{"ja":"Shin'ichiro Omachi","en":""},{"ja":"Hirotomo Aso","en":""}],"title":{"ja":"A Modification of Eigenvalues to Compensate Estimation Errors of Eigenvectors"},"journal":{"ja":"Proceedings 15th International Conference on Pattern Recognition (ICPR2000)"},"volume":"2","pages":"378-381","year":2000,"month":9,"location":"Barcelona, Spain","review":"reviewed","fields":["pattern-recognition"],"doi":"10.1109/ICPR.2000.906091","abstract":{"ja":"In statistical pattern recognition,\nparameters of distributions are\nusually estimated from training samples.\nIt is well known that shortage of training samples causes\nestimation errors which reduce recognition accuracy.\nBy studying estimation errors of eigenvalues,\n various methods of avoiding recognition accuracy reduction\nhave been proposed.\nHowever, estimation errors of eigenvectors have not been considered enough.\nIn this paper, we investigate estimation errors of eigenvectors\n to show these errors are another factor of recognition performance\n reduction.\nWe propose a new method for modifying eigenvalues in order to\nreduce bad influence caused by estimation errors of eigenvectors.\nEffectiveness of the method is shown by experimental results."}},{"key":"Kise2000","id":"836","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Yasuo Miki","en":""},{"ja":"Keinosuke Matsumoto","en":""}],"title":{"ja":"Backgrounds as Information Carriers for Printed Documents"},"journal":{"ja":"Proceedings 15th International Conference on Pattern Recognition (ICPR2000)"},"volume":"4","pages":"380-384","year":2000,"month":9,"review":"reviewed","fields":[]},{"key":"Iwamura2000_MThesis","id":"163","type":"master_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"統計的パターン認識における分布パラメータの推定法に関する研究","en":"A Study on Estimation of Distribution Parameters in Statistical Pattern Recognition"},"year":2000,"month":3,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Mizuno1999","id":"914","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"水野 浩之","en":"Hiroyuki Mizuno"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"松本 啓之亮","en":"Keinosuke Matsumoto"}],"title":{"ja":"単語の出現密度と偏出度を用いた図表と説明テキストの対応付け","en":"Linking Figures and Tables to Their Expository Texts Using Word Density Distributions and Their Biases"},"journal":{"ja":"情報処理学会論文誌","en":"Trans. IPSJ"},"volume":"40","number":"12","pages":"4400-4403","year":1999,"month":12,"review":"reviewed","fields":[]},{"key":"Ohno1999","id":"913","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"大野 潮満","en":"Shiomi Ohno"},{"ja":"前田 英巳子","en":"Emiko Maeda"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"松本 啓之亮","en":"Keinosuke Matsumoto"}],"title":{"ja":"HTMLファイルからのトピック抽出に基づく興味推定とWWW検索","en":"Estimation of User's Interests and WWW Retrieval Based on Topic Extraction from HTML Files"},"journal":{"ja":"電気学会論文誌（Ｃ）","en":"Trans. IEE of Japan"},"volume":"119-C","number":"11","pages":"1316-1322","year":1999,"month":11,"review":"reviewed","fields":[]},{"key":"Iwata1999","id":"904","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"岩田 基","en":"Motoi Iwata"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"松本 啓之亮","en":"Keinosuke Matsumoto"}],"title":{"ja":"隣接グラフを用いた欧文文書画像からの文字列抽出","en":"Text-line Extraction from English Document Images Using the Neighbor Graph"},"journal":{"ja":"情報処理学会論文誌","en":"Trans. IPSJ"},"volume":"40","number":"8","pages":"3239- 3248","year":1999,"month":8,"review":"reviewed","fields":[]},{"key":"Kise1999c","id":"905","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"Keinosuke Matsumoto","en":""}],"title":{"ja":"Text-line Extraction as Selection of Paths in the Neighbor Graph"},"journal":{"ja":"Document Analysis Systems: Theory and Practice, Lecture Notes in Computer Science"},"volume":"1655","pages":"225-239","year":1999,"month":12,"review":"reviewed","fields":[]},{"key":"Kise1999_DAS","id":"833","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"Keinosuke Matsumoto","en":""}],"title":{"ja":"Text-Line Extraction as Selection of Paths in the Neighbor Graph"},"journal":{"ja":"Lecture Notes in Computer Science (Third IAPR Workshop, DAS'98)"},"volume":"1655","pages":"225-239","year":1999,"month":12,"review":"reviewed","fields":[]},{"key":"Kise1999b","id":"950","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Hiroyuki Mizuno","en":""},{"ja":"Masashi Yamaguchi","en":""},{"ja":"Keinosuke Matsumoto","en":""}],"title":{"ja":"On the Use of Density Distributions of Keywords for Automated Generation of Hypertext Links from Arbitrary Parts of Documents"},"journal":{"ja":"Proc. of the 5th International Conference on Document Analysis and Recognition"},"pages":"301-304","year":1999,"month":9,"review":"reviewed","fields":[]},{"key":"Kise1999a","id":"938","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Keinosuke Matsumoto","en":""}],"title":{"ja":"On the Application of Voronoi Diagrams to Page Segmentation"},"journal":{"ja":"Proc. of the Workshop on Document Layout Interpretation and Its Applications"},"number":"IV-C","pages":"1-4","year":1999,"month":9,"review":"reviewed","fields":[]},{"key":"Iwamura1999","id":"56","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shin'ichiro Omachi"},{"ja":"阿曽 弘具","en":"Hirotomo Aso"}],"title":{"ja":"固有ベクトルのずれの影響を軽減する固有値補正法","en":"A Modification Method of Eigenvalues to Avoid Recognition Accuracy Reduction Caused by Estimation Errors of Eigenvectors"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"99","number":"449","presen_id":"PRMU99-146","pages":"93-100","year":1999,"month":11,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"マハラノビス距離などの統計的手法を用いたパターン認識では\nパターンの分布が既知であることが求められるが，一般には未知であり，\n学習サンプルから推定された標本共分散行列を用いる．\nこのとき推定に用いる学習サンプルが少ないと推定誤差が生じ，\n認識性能が低下することが知られている．\n固有値の推定誤差についてはよく調査され，\n固有値を修正する方法が提案されているが，\n固有ベクトルの推定誤差については\nほとんど考慮されてこなかった．\n本論文では，固有ベクトルの推定誤差について調査し，\n固有ベクトルの推定誤差についても認識性能に\n悪影響を与える可能性があることを示す．\nそして，固有ベクトルの推定誤差の影響を\n固有値の修正により軽減する手法を提案する．","en":"Statistical pattern recognition methods such as Mahalanobis\ndistance need to know true distributions of the patterns,\nbut it is usually impossible.\nAs a result, the covariance matrices estimated\nfrom training samples are used.\nIf training samples are not enough,\nestimation errors in both eigenvalues and eigenvectors\nwill occur and they will cause recognition accuracy reduction.\nMany researchers have studied on estimation errors of eigenvalues\nand have proposed various methods to modify eigenvalues.\nHowever, estimation errors of eigenvectors have not been considered enough.\nIn this paper, we investigate estimation errors of eigenvectors\nand show the possibility that they make recognition performance\ndown.\nWe propose a new method estimating eigenvalues so as to\nreduce bad influence on estimation errors of eigenvectors."}},{"key":"Kise1998b","id":"837","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Akinori Sato","en":""},{"ja":"Motoi Iwata","en":""}],"title":{"ja":"Segmentation of page images using the area Voronoi diagram"},"journal":{"ja":"Computer Vision and Image Understanding"},"volume":"70","number":"3","pages":"370-382","year":1998,"month":6,"review":"reviewed","fields":[]},{"key":"Kise1998_IAPR","id":"937","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Motoi Iwata","en":""},{"ja":"Andreas Dengel","en":""},{"ja":"Keinosuke Matsumoto","en":""}],"title":{"ja":"A Computational Geometric Approach to Text-line Extraction from Binary Document Images"},"journal":{"ja":"Proc. of the Third IAPR Workshop on Document Analysis Systems"},"pages":"346-355","year":1998,"month":11,"review":"reviewed","fields":[]},{"key":"Iwamura1998_IEICEPRMU","id":"61","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shin'ichiro Omachi"},{"ja":"阿曽 弘具","en":"Hirotomo Aso"}],"title":{"ja":"クラスタ間情報に基づくマハラノビス距離による文字認識","en":"Character Recognition with Mahalanobis Distance Based on Between-Cluster Information"},"journal":{"ja":"電子情報通信学会技術研究報告","en":"IEICE Technical Report"},"volume":"98","number":"490","presen_id":"PRMU98-156","pages":"49-54","year":1998,"month":12,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Iwamura1998_Conf","id":"63","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"},{"ja":"大町 真一郎","en":"Shin'ichiro Omachi"},{"ja":"阿曽 弘具","en":"Hirotomo Aso"}],"title":{"ja":"認識率に寄与する文字画像の固有ベクトル"},"journal":{"ja":"平成10年度電気関係学会東北支部連合大会","en":"1998 Tohoku-Section Joint Convention Record of Institutes of Electrical and Information Engineers"},"presen_id":"2G-24","pages":"286","year":1998,"month":8,"review":"not_reviewed","fields":["pattern-recognition"],"abstract":{"ja":"部分空間法は少数の低次の固有ベクトルのみを用いて認識を行うものであり，\nマハラノビス距離を用いた手法でも計算量削減のために低次の固有値・固有ベクトル\nのみを用いた認識手法等が提案されている．\nその場合，認識に有効な固有ベクトルを選択することが必要となるが，\nこれまで各固有ベクトルが認識率にどのように寄与するかの考察は\nほとんど行われていなかった．\n\n本研究では，マハラノビス距離および部分空間法の類似度において認識に寄与する\n固有ベクトルを考察するために，一部の固有ベクトルのみを用いた\n認識実験を行った．\nその結果，マハラノビス距離では比較的高次の固有ベクトルが，\n部分空間法の類似度では低次の固有ベクトルのみが認識に寄与していることが\n分かった．\nまた，部分空間法が少ない次元数で高精度な認識を\n実現できることが確認された．"}},{"key":"Iwata1998","id":"936","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"岩田 基","en":""},{"ja":"黄瀬 浩一","en":""},{"ja":"松本 啓之亮","en":""}],"title":{"ja":"隣接グラフによる文字列抽出"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU'98)講演論文集"},"volume":"?","pages":"281-286","year":1998,"month":7,"review":"not_reviewed","fields":[]},{"key":"Iwamura1998_BThesis","id":"162","type":"bachelors_thesis","language":"JA","show":"show_except_datafile","authors":[{"ja":"岩村 雅一","en":"Masakazu Iwamura"}],"title":{"ja":"文字画像の固有ベクトル成分を活用する文字認識に関する研究"},"year":1998,"month":3,"review":"not_reviewed","fields":["pattern-recognition"]},{"key":"Kise1998_PRMU","id":"941","type":"survey","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":""}],"title":{"ja":"文書画像理解の目指すもの"},"journal":{"ja":"電子情報通信学会技術研究報告，パターン認識・メディア理解研究会 (PRMU97-248)"},"pages":"55-62","year":1998,"month":2,"review":"not_reviewed","fields":[],"note":{"ja":"特別講演"}},{"key":"Kise1998a","id":"834","type":"other","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Osamu Yanagida","en":""}],"title":{"ja":"Page segmentation using thinning of white areas"},"journal":{"ja":"Systems and Computers in Japan"},"volume":"29","number":"3","pages":"59-68","year":1998,"month":3,"review":"not_reviewed","fields":[]},{"key":"Kise1997_IEICE","id":"949","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"黄瀬浩一","en":"Koichi Kise"},{"ja":"柳田 修","en":"Osamu Yanagida"}],"title":{"ja":"白領域の細線化表現を用いた文書画像の領域分割","en":"Page Segmentation Using Thinning of White Areas"},"journal":{"ja":"電子情報通信学会論文誌","en":"Trans. of IEICE"},"volume":"J80-D-II","number":"6","pages":"1608-1616","year":1997,"month":6,"review":"reviewed","fields":[]},{"key":"Izumi1997","id":"910","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"和泉 憲明","en":"Noriaki Izumi"},{"ja":"高松 忍","en":"Shinobu Takamatsu"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"福永 邦雄","en":"Kunio Fukunaga"}],"title":{"ja":"プロセス代数と様相論理を融合した並行システムの仕様記述","en":"A Specification Description of Concurrent Systems Using Process Algebra and Modal Logic"},"journal":{"ja":"電子情報通信学会論文誌","en":"Trans. IEICE"},"volume":"J80-D-I","number":"4","pages":"411-414","year":1997,"month":4,"review":"reviewed","fields":[]},{"key":"Kise1997_IEEE","id":"935","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Akinori Sato","en":""},{"ja":"Keinosuke Matsumoto","en":""}],"title":{"ja":"Document Image Segmentation as Selection of Voronoi Edges"},"journal":{"ja":"Proc. of the IEEE Workshop on Document Image Analysis"},"pages":"32-39","year":1997,"month":6,"review":"reviewed","fields":[]},{"key":"Kise1997","id":"940","type":"survey","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":""},{"ja":"小島 篤博","en":""},{"ja":"福永 邦雄","en":""}],"title":{"ja":"画像・図と言語の統合"},"book_title":"システム／制御／情報","volume":"41","number":"12","pages":"520-527","year":1997,"month":12,"review":"not_reviewed","fields":[]},{"key":"Tani1996","id":"909","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"谷 忠明","en":"Tadaaki Tani"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"高松 忍","en":"Shinobu Takamatsu"},{"ja":"福永 邦雄","en":"Kunio Fukunaga"}],"title":{"ja":"解釈の曖昧性を考慮した図と説明テキストの統合理解","en":"Understanding of Diagrams and Expository Text by Resolving Ambiguities of Interpretation"},"journal":{"ja":"電気学会論文誌（Ｃ）","en":"Trans. IEE of Japan"},"volume":"116-C","number":"11","pages":"1202-1210","year":1996,"month":11,"review":"reviewed","fields":[]},{"key":"Kise1996_IEE","id":"900","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"秋吉 裕二","en":"Yuji Akiyoshi"},{"ja":"高松 忍","en":"Shinobu Takamatsu"}],"title":{"ja":"白領域の有効面積に基づく文書画像領域分割","en":"Page Segmentation Based on Effective Area of Background"},"journal":{"ja":"電気学会論文誌（Ｃ）","en":"Trans. IEE of Japan"},"volume":"116-C","number":"9","pages":"1035-1042","year":1996,"month":9,"review":"reviewed","fields":[]},{"key":"Izumi1996","id":"934","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"N. Izumi","en":""},{"ja":"S. Takamatsu","en":""},{"ja":"K. Kise","en":""},{"ja":"K. Fukunaga","en":""}],"title":{"ja":"A First-Order Theory of Agent Models with a Layered Architecture"},"journal":{"ja":"Proc. of the Second International Conference on Multiagent Systems"},"pages":"444","year":1996,"month":12,"review":"reviewed","fields":[]},{"key":"Kise1996b","id":"899","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Osamu Yanagida","en":""},{"ja":"Shinobu Takamatsu","en":""}],"title":{"ja":"Page Segmentation Based on Thinning of Background"},"journal":{"ja":"Proc. of the 13th International Conference on Pattern Recognition"},"volume":"3","pages":"788-792","year":1996,"month":8,"review":"reviewed","fields":[]},{"key":"Izumi1996_IASTED","id":"933","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"N. Izumi","en":""},{"ja":"S. Takamatsu","en":""},{"ja":"K. Kise","en":""},{"ja":"K. Fukunaga","en":""}],"title":{"ja":"A First-Order Logical Formalization of Multi-Agent Models with Process Algebraic Formulation"},"journal":{"ja":"Proc. of the IASTED International Conference on Artificial Intelligence, Expert Systems and Neural Networks"},"pages":"116-119","year":1996,"month":8,"review":"reviewed","fields":[]},{"key":"Kojima1996","id":"932","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"A. Kojima","en":""},{"ja":"N. Izumi","en":""},{"ja":"S. Takamatsu","en":""},{"ja":"K. Kise","en":""},{"ja":"K. Fukunaga","en":""}],"title":{"ja":"A Personal Agent Dealing with Situation-Dependent Tasks"},"journal":{"ja":"Proc. of the IASTED International Conference on Artificial Intelligence, Expert Systems and Neural Networks"},"pages":"33-35","year":1996,"month":8,"review":"reviewed","fields":[]},{"key":"Kise1996_MIRU","id":"930","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":""},{"ja":"柳田 修","en":""},{"ja":"高松 忍","en":""}],"title":{"ja":"白領域の細線化を用いた文書画像の領域分割"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU'96)講演論文集"},"volume":"?","pages":"337-342","year":1996,"month":7,"review":"not_reviewed","fields":[]},{"key":"Oohiraki1996","id":"931","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"大開 保","en":""},{"ja":"黄瀬 浩一","en":""},{"ja":"高松 忍","en":""}],"title":{"ja":"電子図書館構築のための印刷文書のハイパーテキスト化"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU'96)講演論文集"},"volume":"?","pages":"277-282","year":1996,"month":7,"review":"not_reviewed","fields":[]},{"key":"Kise1996a","id":"897","type":"other","language":"EN","show":"show","authors":[{"ja":"K. Kise","en":""},{"ja":"T. Shiraishi","en":""},{"ja":"S. Takamatsu","en":""},{"ja":"K. Fukunaga","en":""}],"title":{"ja":"A Method of Post-Processing for Character Recognition Based on Syntactic and Semantic Analysis of Sentences"},"journal":{"ja":"Systems and Computers in Japan"},"volume":"27","number":"9","pages":"94-107","year":1996,"month":8,"review":"not_reviewed","fields":[]},{"key":"Kise1995_IPSJ","id":"898","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"矢島 尚子","en":"Naoko Yajima"},{"ja":"高松 忍","en":"Shinobu Takamatsu"},{"ja":"福永 邦雄","en":"Kunio Fukunaga"}],"title":{"ja":"文書画像構造解析のためのインクリメンタルな知識獲得","en":"Incremental Knowledge Acquisition for Structure Analysis of Document Images"},"journal":{"ja":"情報処理学会論文誌","en":"Trans. IPSJ"},"volume":"36","number":"11","pages":"2622-2631","year":1995,"month":11,"review":"reviewed","fields":[]},{"key":"Hattori1995","id":"911","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"服部 洋一","en":"Yoichi Hattori"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"北橋 忠宏","en":"Tadahiro Kitahashi"},{"ja":"福永 邦雄","en":"Kunio Fukunaga"}],"title":{"ja":"動的機能のモデルに基づく物体の機能認識","en":"Recognition of Functions of Objects Based on Models of Dynamic Functions"},"journal":{"ja":"情報処理学会論文誌","en":"Trans. IPSJ"},"volume":"36","number":"10","pages":"2277-2285","year":1995,"month":10,"review":"reviewed","fields":[]},{"key":"Takamatsu1995","id":"908","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"高松 忍","en":"Shinobu Takamatsu"},{"ja":"和泉 憲明","en":"Noriaki Izumi"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"福永 邦雄","en":"Kunio Fukunaga"}],"title":{"ja":"ハードウェア設計仕様における自然言語記述の一階論理に基づく状況的・動的解釈","en":"First-Order Logic Based Situational and Dynamic Interpretation of Natural Language Descriptions in Hardware Design Specifications"},"journal":{"ja":"人工知能学会論文誌","en":"Trans. of JSAI"},"volume":"10","number":"5","pages":"720-730","year":1995,"month":9,"review":"reviewed","fields":[]},{"key":"Kise1995_ICDAR","id":"832","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Noriyoshi Yoneda","en":""},{"ja":"Shinobu Takamatsu","en":""},{"ja":"Kunio Fukunaga","en":""}],"title":{"ja":"Interpretation of conceptual diagrams from line segments and strings"},"journal":{"ja":"Proc. 3rd International Conference on Document Analysis and Recognition (ICDAR1995)"},"pages":"960-963","year":1995,"month":8,"review":"reviewed","fields":[]},{"key":"Kise1994_IEICE_Method","id":"896","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"白石 忠道","en":"Tadamichi Shiraishi"},{"ja":"高松 忍","en":"Shinobu Takamatsu"},{"ja":"福永 邦雄","en":"Kunio Fukunaga"}],"title":{"ja":"構文・意味解析を用いた文字認識後処理法","en":"A Method of Post-Processing for Character Recognition Based on Syntactic and Semantic Analysis of Sentences"},"journal":{"ja":"電子情報通信学会論文誌","en":"Trans. IEICE"},"volume":"J77-D-II","number":"11","pages":"2199-2209","year":1994,"month":11,"review":"reviewed","fields":[]},{"key":"Kise1994_IEE","id":"906","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"白石 忠道","en":"Tadamichi Shiraishi"},{"ja":"高松 忍","en":"Shinobu Takamatsu"},{"ja":"福永 邦雄","en":"Kunio Fukunaga"}],"title":{"ja":"幾何問題の理解と解決を目的とした図と文の統合理解","en":"Integrated Understanding of Sentences and Diagrams for Understanding and Solution of Geometry Problems"},"journal":{"ja":"電気学会論文誌（Ｃ）","en":"Trans. IEE of Japan"},"volume":"114-C","number":"11","pages":"1201-1207","year":1994,"month":11,"review":"reviewed","fields":[]},{"key":"Kise1994_IEICE","id":"895","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Noboru Babaguchi","en":""}],"title":{"ja":"Representing, Utilizing and Acquiring Knowledge for Document Image Understanding"},"journal":{"ja":"IEICE Transactions on Information and Systems"},"volume":"E77-D","number":"7","pages":"770-777","year":1994,"month":7,"review":"reviewed","fields":[]},{"key":"Yoneda1994","id":"929","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Noriyoshi Yoneda","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Shinobu Takamatsu","en":""},{"ja":"Kunio Fukunaga","en":""}],"title":{"ja":"A Method of Understanding Conceptual Diagrams"},"journal":{"ja":"Proc. of the IAPR Workshop on Machine Vision Applications '94"},"pages":"334-337","year":1994,"month":12,"review":"reviewed","fields":[]},{"key":"Hattori1994","id":"928","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"服部 洋一","en":""},{"ja":"秋山 文人","en":""},{"ja":"黄瀬 浩一","en":""},{"ja":"北橋 忠宏","en":""},{"ja":"高松 忍","en":""},{"ja":"福永 邦雄","en":""}],"title":{"ja":"力の入出力に着目した道具の機能モデル表現と物体認識への適用"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU'94)講演論文集"},"volume":"?","pages":"1-8","year":1994,"month":7,"review":"not_reviewed","fields":[]},{"key":"Kise1993_IPSJ_Hypothesis","id":"893","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"百田 賢一","en":"Ken-ichi Momota"},{"ja":"杉山 淳一","en":"Jun-ichi Sugiyama"},{"ja":"馬場口 登","en":"Noboru Babaguchi"},{"ja":"手塚 慶一","en":"Yoshikazu Tezuka"}],"title":{"ja":"レイアウトとコンテントの知識を用いた仮説駆動型文書画像理解","en":"Hypothesis Driven Document Image Understanding with Knowledge about Layout and Content"},"journal":{"ja":"情報処理学会論文誌","en":"Trans. IPSJ"},"volume":"34","number":"8","pages":"1716-1730","year":1993,"month":8,"review":"reviewed","fields":[]},{"key":"Kise1993_IPSJ_Organizing","id":"892","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"山岡 正輝","en":"Masaki Yamaoka"},{"ja":"馬場口 登","en":"Noboru Babaguchi"},{"ja":"手塚 慶一","en":"Yoshikazu Tezuka"}],"title":{"ja":"文書画像構造解析のための知識ベースの一構成法","en":"Organizing Knowledge-Base for Structure Analysis of Document Images"},"journal":{"ja":"情報処理学会論文誌","en":"Trans. IPSJ"},"volume":"34","number":"1","pages":"75-87","year":1993,"month":1,"review":"reviewed","fields":[]},{"key":"Kise1993b","id":"927","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Yoichi Hattori","en":""},{"ja":"Tadahiro Kitahashi","en":""},{"ja":"Kunio Fukunaga","en":""}],"title":{"ja":"Representing and Recognizing Simple Hand-tools Based on their Functions"},"journal":{"ja":"Proc. of the First Asian Conference on Computer Vision"},"pages":"656-659","year":1993,"month":11,"review":"reviewed","fields":[]},{"key":"Kise1993a","id":"894","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Naoko Yajima","en":""},{"ja":"Noboru Babaguchi","en":""},{"ja":"Kunio Fukunaga","en":""}],"title":{"ja":"Incremental Acquisition of Knowledge about Layout Structures from Examples of Documents"},"journal":{"ja":"Proc. of the Second International Conference on Document Analysis and Recognition"},"pages":"668-　671","year":1993,"month":10,"review":"reviewed","fields":[]},{"key":"Kise1992_IAPR","id":"926","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Tadaaki Shiraishi","en":""},{"ja":"Shinobu Takamatsu","en":""},{"ja":"Hiroji Kusaka","en":""}],"title":{"ja":"Improvement of Text Image Recognition Based on Linguistic Constraints"},"journal":{"ja":"Proc. of the IAPR Workshop on Machine Vision Applications '92"},"pages":"511-514","year":1992,"month":12,"review":"reviewed","fields":[]},{"key":"Kise1992","id":"891","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Masaki Yamaoka","en":""},{"ja":"Noboru Babaguchi","en":""},{"ja":"Yoshikazu Tezuka","en":""}],"title":{"ja":"Model Based System for Analyzing Document Images"},"journal":{"ja":"Proc. of the 11th International Conference on Pattern Recognition"},"volume":"2","pages":"647-650","year":1992,"month":9,"review":"reviewed","fields":[]},{"key":"Kise1992_MIRU","id":"925","type":"domestic_conf","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":""},{"ja":"坂東 博司","en":""},{"ja":"高松 忍","en":""},{"ja":"日下 浩次","en":""}],"title":{"ja":"文書構造解析のためのレイアウト構造のクラスタリング"},"journal":{"ja":"画像の認識・理解シンポジウム(MIRU'92)講演論文集"},"volume":"1","pages":"287-294","year":1992,"month":7,"review":"not_reviewed","fields":[]},{"key":"Babaguchi1991b","id":"890","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Noboru Babaguchi","en":""},{"ja":"Koji Yamada","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Yoshikazu Tezuka","en":""}],"title":{"ja":"Connectionist Model Binarization"},"journal":{"ja":"Neural Networks in Pat-tern Recognition and Their Applications, World Scientific"},"pages":"127-142","year":1991,"month":12,"review":"reviewed","fields":[]},{"key":"Babaguchi1991a","id":"889","type":"journal_paper","language":"EN","show":"show","authors":[{"ja":"Noboru Babaguchi","en":""},{"ja":"Koji Yamada","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Yoshikazu Tezuka","en":""}],"title":{"ja":"Connectionist Model Binarization"},"journal":{"ja":"International Journal of Pattern Recognition and Artificial Intelligence"},"volume":"5","number":"4","pages":"629-644","year":1991,"month":10,"review":"reviewed","fields":[]},{"key":"Babaguchi1991_J","id":"888","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"馬場口 登","en":""},{"ja":"黄瀬 浩一","en":""},{"ja":"手塚 慶一","en":""}],"title":{"ja":"ニューラルネットを用いた画像２値化"},"journal":{"ja":"日本医用画像工学会誌"},"volume":"9","number":"4","pages":"410-415","year":1991,"month":9,"review":"reviewed","fields":[]},{"key":"Babaguchi1990_IEICE","id":"887","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"馬場口 登","en":"Noboru Babaguchi"},{"ja":"山田 耕児","en":"Koji Yamada"},{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"手塚 慶一","en":"Yoshikazu Tezuka"}],"title":{"ja":"コネクショニストモデルによる画像２値化の実験的検討","en":"An Experimental Consideration on Image Binarization by Connectionist Model"},"journal":{"ja":"電子情報通信学会論文誌","en":"Trans. IEICE"},"volume":"J73-D-II","number":"8","pages":"1281-1287","year":1990,"month":8,"review":"reviewed","fields":[]},{"key":"Kise1990","id":"924","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Jun-ichi Sugiyama","en":""},{"ja":"Masaki Yamaoka","en":""},{"ja":"Ken-ichi Momota","en":""},{"ja":"Noboru Babaguchi","en":""},{"ja":"Yoshikazu Tezuka","en":""}],"title":{"ja":"Model Based Understanding of Document Images"},"journal":{"ja":"Proc. of the IAPR Workshop on Machine Vision Applications '90"},"pages":"471-474","year":1990,"month":11,"review":"reviewed","fields":[]},{"key":"Babaguchi1990","id":"886","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Noboru Babaguchi","en":""},{"ja":"Koji Yamada","en":""},{"ja":"Koichi Kise","en":""},{"ja":"Yoshikazu Tezuka","en":""}],"title":{"ja":"Connectionist Model Binarization"},"journal":{"ja":"Proc. of the 10th International Conference on Pattern Recognition"},"pages":"51-56","year":1990,"month":6,"review":"reviewed","fields":[]},{"key":"Kise1989","id":"885","type":"journal_paper","language":"JA","show":"show","authors":[{"ja":"黄瀬 浩一","en":"Koichi Kise"},{"ja":"杉山 淳一","en":"Jun-ichi Sugiyama"},{"ja":"馬場口 登","en":"Noboru Babaguchi"},{"ja":"手塚 慶一","en":"Yoshikazu Tezuka"}],"title":{"ja":"レイアウトモデルに基づく文書構造解析","en":"Layout Model Based Analysis of Document Structure"},"journal":{"ja":"電子情報通信学会論文誌","en":"Trans. IEICE"},"volume":"J72-D-II","number":"7","pages":"1029-1039","year":1989,"month":7,"review":"reviewed","fields":[]},{"key":"Kise1988","id":"884","type":"intl_conf","language":"EN","show":"show","authors":[{"ja":"Koichi Kise","en":""},{"ja":"Koji Yamada","en":""},{"ja":"Naoki Tanaka","en":""},{"ja":"Noboru Babaguchi","en":""},{"ja":"Yoshikazu Tezuka","en":""}],"title":{"ja":"Visiting Card Understanding System"},"journal":{"ja":"Proc. of the 9th International Conference on Pattern Recognition"},"pages":"425-429","year":1988,"month":11,"review":"reviewed","fields":[]}]}