チャーノフの顔形グラフにおける顔形要素の評価
スポンサーリンク
概要
- 論文の詳細を見る
Face graph of representing multivariate data introduced by H. Chernoff^2 consists of the cartoon of a face which is drawn by the computer and determined by 18 parameters (features) including length of nose, curvature of mouth, slant of eyes, etc. He pointed out^3 that discussions of this approach typically raise the following questions; Which features should components of observation vector be assigned? How would the results of this subjective approach be affected by a permutation of the components of vector? Which features are most effective in communicating information in this context? In this study, the experiments by an orthogonal array table was designed to evaluate the conjecture that the observer's reaction to a facial expression will quickly determine the relevant information with negligible dependence on the choice of features. After the variation of features was decomposed into four terms: the linear term (main effect), a part of non-linear term (interaction effect), other non-linear term (error of experiment) and the observer's error of reaction, these factors were assigned to L_<32>(2^<31>) orthogonal array table and also the experiments of the observer's visual reaction were carried out by applying the semantic differential (SD) method. The results are as follows. (1) The contribution of the effects of these factors to a pair of adjective "self-confident vs. self- distrustful (laughing vs. weeping)" which show the highest (lowest) reaction are 41.7(23.5)%, 5.3(4.0)%, 1.8(1.9)% and 53(70.8)% respectively. In consequence, the observer's visual reaction to face graph is expressed by the linear combinatios of variation of features though reaction error is larger than 50%. (2) The effective features to visual reaction are slant of eyebrows, curvature of mouth, slant of eyes and interaction between slant of eyebrows and curvature of mouth in sequence. Accordingingly, it is certified that the eyes talk about as much as the mouth does.
- 明治大学の論文
著者
関連論文
- 2-6 チャーノフの顏形グラフの応用研究(1)(第16回年次大会)
- 1-3 箱ひげ管理図の研究(1)(第16回年次大会)
- 点訳システムに関する研究 : 漢字仮名混じり文処理について
- フールプルーフ化を考慮した教育用ロボット言語の開発
- 教育用ロボット言語の開発の開発-2-ALARM ver.2.0の開発
- チャ-ノフ顔形グラフの応用研究-2-大気汚染の地域特性の解析
- チャーノフの顔形グラフの応用研究(II) : 大気汚染の地域特性の解析
- チャーノフの顔形グラフの応用研究(I)
- 点字ワードプロセッサに関する研究
- チャ-ノフの顔形グラフの顔形要素に対する個人誤差の評価-2-
- チャーノフの顔形グラフの顔形要素に対する個人誤差の評価(II)
- 大気汚染物質の分布パターンに関する研究
- パーソナルコンピュータを利用した点字ワードプロセッサシステムの研究
- 3-8 顔形グラフの研究(1) : 顔形要素の評価(第25回 研究発表会)
- 教育用ロボット言語の開発
- チャーノフの顔形グラフの顔形要素に対する個人誤差の評価
- チャーノフの顔形グラフにおける顔形要素の評価
- カナ・点字変換システムの研究
- ファッション画におけるディフォルマシオンの定量的評価(II) : 多元配置法による視覚感度実験(第34回研究発表大会)
- ファッション画におけるディフォルマシオンの定量的評価(I) : 直交配列表による視覚感度実験(第34回研究発表大会)