Color-space analysis for facial feature detection

Abstract

Facial feature detection is an important preprocessing step used in facial image processing. Most of the image-based facial feature detection techniques use only the luminance information ignoring the color information of the images. We introduce a method of adding color-space information to the principal component analysis (PCA) based eigen features technique. The experimental results show that the color information helps to improve the detection performance of the detectors when compared with the detectors using only the luminance information. © 2002 Croatian Soc. Electronics in Marine-ELMAR.