Converting facial expressions using recognition-based analysis of image sequences

Takahiro Otsuka, Jun Ohya

研究成果: Conference contribution

抜粋

A method for converting one person’s facial expression into another person’s is proposed. The sequence of the feature vector for each expression is modeled by using HMM with the hidden states corresponding to the different muscle conditions (relaxed, contracting, and the end of contraction). The probability of each state is evaluated for each frame and the contraction rate of each muscle is obtained from the probability of each state using a matrix representing the characteristics of other people’s expressions. The experiments showed the superior realism of the expression generated by our proposed method.

元の言語English
ホスト出版物のタイトルComputer Vision - ACCV 1998 - 3rd Asian Conference on Computer Vision, Proceedings
編集者Roland Chin, Ting-Chuen Pong
出版者Springer Verlag
ページ703-710
ページ数8
ISBN(印刷物)3540639314, 9783540639312
DOI
出版物ステータスPublished - 1997 1 1
外部発表Yes
イベント3rd Asian Conference on Computer Vision, ACCV 1998 - Hong Kong, Hong Kong
継続期間: 1998 1 81998 1 10

出版物シリーズ

名前Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
1352
ISSN(印刷物)0302-9743
ISSN(電子版)1611-3349

Other

Other3rd Asian Conference on Computer Vision, ACCV 1998
Hong Kong
Hong Kong
期間98/1/898/1/10

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

フィンガープリント Converting facial expressions using recognition-based analysis of image sequences' の研究トピックを掘り下げます。これらはともに一意のフィンガープリントを構成します。

  • これを引用

    Otsuka, T., & Ohya, J. (1997). Converting facial expressions using recognition-based analysis of image sequences. : R. Chin, & T-C. Pong (版), Computer Vision - ACCV 1998 - 3rd Asian Conference on Computer Vision, Proceedings (pp. 703-710). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); 巻数 1352). Springer Verlag. https://doi.org/10.1007/3-540-63931-4_280