Presentation 2004/9/19
Extraction of facial regions from images using Self Organizing Maps
Yoshinori KAWASE, Tomoharu NAGAO,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) This report presents a method to extract facial regions from human images. As one of such methods, a method to categorize each rectangular region in a give image into facial or not facial is known. Template matching method is a typical one, and each rectangular region is compared with a standard template image by calculating a distance between them. The most typical distance is root mean square of differences between gray values of the template image pixels and the corresponding pixels of the rectangular region. In this case, a lot of error detections will occur since two dimensional distribution of pixel values is not considered. Therefore, we propose to use SOM; Self Organizing Maps known as one of the most useful classifying method without teachers. First, we construct an one dimensional map to categorize a rectangular region into facial or not facial through learning. Secondly, we show that the detection method using this map is more useful than template matching through several experimental results.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Facial Regions / SOM / Self Organizing Maps / Template Matching
Paper # HCS2004-21
Date of Issue

Conference Information
Committee HCS
Conference Date 2004/9/19(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) Extraction of facial regions from images using Self Organizing Maps
Sub Title (in English)
Keyword(1) Facial Regions
Keyword(2) SOM
Keyword(3) Self Organizing Maps
Keyword(4) Template Matching
1st Author's Name Yoshinori KAWASE
1st Author's Affiliation Graduate School of Environment and Information Sciences()
2nd Author's Name Tomoharu NAGAO
2nd Author's Affiliation Graduate School of Environment and Information Sciences
Date 2004/9/19
Paper # HCS2004-21
Volume (vol) vol.104
Number (no) 320
Page pp.pp.-
#Pages 6
Date of Issue