IPIU 1998 (제10회 영상처리 및 이해에 관한 워크샵)
Current Result Document :
| 한글제목(Korean Title) |
Facial Feature Point Tracking Using Moving Color |
| 영문제목(English Title) |
Facial Feature Point Tracking Using Moving Color |
| 저자(Author) |
Nam-Ho Kim
Sang-Jin Ham
Sang Chul Ahn
Hyuong-Gon Kim
Sung-Jea Ko
|
| 원문수록처(Citation) |
VOL 10 NO. 01 PP. 0127 ~ 0132 (1998. 01) |
한글내용
(Korean Abstract) |
|
영문내용
(English Abstract) |
This paper presents an automatic facial feature point extraction algorithm in sequential color video images. To extract facial region in a video sequence, a moving color detection technique is proposed that emphasizes moving skin color region by applying a motion detection algorithm to skin-color transform based on GFCD(generalized face color distribution) is proposed. For the motion detection, AWUPC(adaptive weighted unmatched pixel count) method is proposed. The threshold for AWUPC is adaptively changed according to color transformed pixel values that represent the probability of the desired color information. An eye candidate region is extracted using both of the black/white color transform of original images and the valley region detected using morphological operations from the moving skin color region. Facial components such eyes and mouth is extracted from the geometrical relationships and histogram of color transformed image. To extract the exact feature points, the principal component analysis is used on each eye and mouth regions. Experimental results show that the feature points of eyes and mouth can be correctly obtained irrespectively of background, individual variation, face orientation and size. And, these methods were implemented in a specific image processing system(TMS320C80) for real-time applications.
|
| 키워드(Keyword) |
Facial feature extraction
Motion detection
Color model
Moving color
|
| 파일첨부 |
PDF 다운로드
|
