Comparison of Video Face Detection methods Using HSV, HSL and HSI Color Spaces

Abstract

this paper presents, new face detection methods based on HSL and HSI color spaces are presented. A comparison of the new face detection methods and a new HSV skin color range is presented. The three color spaces are based on: H, S, V, L, and I, whose represent Hue, Saturation, Value, Luminance and Intensity respectively. YouTube Celebrities Face Tracking and Recognition Dataset is used. It contains 1910 sequences of 47 subjects. All dataset videos are encoded in MPEG4 at 25fps rate. The proposed methods based on two main steps, at the beginning, the skin like regions is detected by the gradient values of the proposed color space. According to main facial features, such as eyes, mouth and nose the desired faces are determined from the recommended regions. According to experimental results, HSV color space gives good results in lighten_faces, HSL color space gives good results for multi_faces and HSI color space gives good results for single_faces and zoomed_faces videos.