@techreport{oai:ipsj.ixsq.nii.ac.jp:00080145,
 author = {西村, 朋己 and 呉, 海元 and 瀧, 寛和 and 三浦, 浩一 and Tomoki, Nishimura and Haiyuan, Wu and Hirokazu, Taki and Hirokazu, Miura},
 issue = {13},
 month = {Jan},
 note = {本稿では，一般物体認識の精度向上の為に，関連成分分析 （RCA: relevant component analysis） を用いた局所特徴変換法を提案する．提案手法では，各々の学習画像セットから得られた局所特徴間の距離・密度に基づき特徴量を変換し各々のカテゴリーに顕著な特徴を作成する．今回は，生活支援システムへの応用を想定し，日常生活用品を認識対象とした共通データベースを用いて，提案手法の有効性を確認する実験を行う．, In this paper, we propose a method for transforming local feature using relevant component analysis (RCA). Our method analyzes the distribution of E-SIFT local features provided from a training image set, and makes the difference of visual word clear by adjusting distance between local features that followed the density of the distribution. In this paper, we do mind an application to life supporting system. This method ’s effectiveness was confirmed through several comparison experiments using common database of everyday life article.},
 title = {RCAを用いた局所特徴変換法と生活支援用一般物体認識への応用},
 year = {2012}
}