Human face detection from color images: preliminary result

Abstract

Images containing faces are essential to intelligent vision-based human computer interaction where it is utilized in various applications (e.g. face recognition, face tracking, pose estimation, expression estimation, etc). This paper aims to provide preliminary investigation on human face detection from a color image. A successful and robust algorithm should capable in identifying all facial region within a given images regardless of its position, orientation and lighting conditions. Such problems are quite challenging because the faces are non-rigid and have a high degree of variability in size, shape, color and texture. Numerous approaches have been developed and proposed to localize facial region in a single image. Our study utilized YCrCb color space method in the detection of human face from a color image with a pre-determine chrominance and luminance parameters. The result obtained illustrates the feasibility of this approach where it exhibits high and accurate segmentation rate though possess minor flaws.