Presentation | 2005/3/17 Quantitative Analysis for Impression Expression of Facial Features And Its Application to Discrimination of Facial Attributes Fumiyasu NISHINO, Masahide KANEKO, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | Facial caricatures represent features of individual faces compactly and effectively, and they are frequently used in the various situations such as newspapers, magazines, web, and so on. Description of facial features using eigenspaces enables not only automatic drawing facial caricatures, but also various quantitative discussions for facial impressions. This paper proposes a system which can output how much each of features of the sex, age and six basic facial expressions (happiness, surprise, fear, anger, disgust, sadness) is contained in an inputted face. Firstly the principal component analysis for facial features using the face database is described. An average face of each of above categories is calculated, and the orthogonal expansion to obtained eigenspaces are carried out for each average face. By this it becomes clear that the distribution of important principal components is different for an average face corresponding to each category. Then, the orthogonal expansion of inputted face into calculated eigenspaces is carried out. How close expansion coefficients of inputted face to those of each of average faces of different categories is distinguished, and the similarity between each principal component is evaluated. This makes it possible to estimate that how much an inputted face contains each of facial impressions such as sex, age, and six basic facial expressions. Obtained similarity values are close to results of human's subjective evaluation. Therefore the proposed method is useful to quantitative analysis of facial impressions. This system can treat various facial impression words by preparing a typical face corresponding to each of facial impression words. In the future, it will be possible to retrieve faces having specific facial impressions from facial database by using the proposed method. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Facial Caricature / Eigenspace / Principal Component Analysis / Facial Impressions / Features in Shape / Features in Arrangement / Six Basic Facial Expressions |
Paper # | HCS2004-51 |
Date of Issue |
Conference Information | |
Committee | HCS |
---|---|
Conference Date | 2005/3/17(1days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | |
Vice Chair | |
Secretary | |
Assistant |
Paper Information | |
Registration To | Human Communication Science (HCS) |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Quantitative Analysis for Impression Expression of Facial Features And Its Application to Discrimination of Facial Attributes |
Sub Title (in English) | |
Keyword(1) | Facial Caricature |
Keyword(2) | Eigenspace |
Keyword(3) | Principal Component Analysis |
Keyword(4) | Facial Impressions |
Keyword(5) | Features in Shape |
Keyword(6) | Features in Arrangement |
Keyword(7) | Six Basic Facial Expressions |
1st Author's Name | Fumiyasu NISHINO |
1st Author's Affiliation | Department of Electronic Engineering, The University of Electro-Communications() |
2nd Author's Name | Masahide KANEKO |
2nd Author's Affiliation | Department of Electronic Engineering, The University of Electro-Communications |
Date | 2005/3/17 |
Paper # | HCS2004-51 |
Volume (vol) | vol.104 |
Number (no) | 744 |
Page | pp.pp.- |
#Pages | 6 |
Date of Issue |