Presentation 1997/7/24
Face Area Segmentation using Partial Features and Face Recognition
Tatsuo SATO, Kiotaka MIURA, Yohko FUKUDA, Hidehiro OHKI, Tsutomu ENDO,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In case of identifying a person using face image recognition by computers, changes of hairstyles, clothes and scenery affects correctness of the result. Segmentation of face area is necessary to avoid this influence. The segmentation is based on positions of face elements such as eyes, nose and mouth. Generally, as shape of these face elements changes variously, it is hard to detect them by conventional template matching in image space. This paper proposes a flexible pattern matching method that measures distance in feature space rather than in image space. The method is based on discriminant analysis and uses local autocorrelation features extended to deal with color images. A method of segmentation of face area is shown as well. We performed some experiments confirming validity of the method. One is to detect some parts of the face elements as partial features. Another is segmentation. And then we performed other two experiments of recognition using original images and the detected rectangle area images, and we compared the recognition rates.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Pattern matching / Partial features / Color image / Face recognition
Paper # MVE97-55
Date of Issue

Conference Information
Committee MVE
Conference Date 1997/7/24(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 Media Experience and Virtual Environment (MVE)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Face Area Segmentation using Partial Features and Face Recognition
Sub Title (in English)
Keyword(1) Pattern matching
Keyword(2) Partial features
Keyword(3) Color image
Keyword(4) Face recognition
1st Author's Name Tatsuo SATO
1st Author's Affiliation Oita Industrial Research Institute()
2nd Author's Name Kiotaka MIURA
2nd Author's Affiliation Faculty of Engineering, Oita University
3rd Author's Name Yohko FUKUDA
3rd Author's Affiliation Faculty of Engineering, Oita University
4th Author's Name Hidehiro OHKI
4th Author's Affiliation Faculty of Engineering, Oita University
5th Author's Name Tsutomu ENDO
5th Author's Affiliation Faculty of Engineering, Oita University
Date 1997/7/24
Paper # MVE97-55
Volume (vol) vol.97
Number (no) 206
Page pp.pp.-
#Pages 8
Date of Issue