An Appropriate Color Space to Improve Human Skin Detection

Abstract

Skin color detection is often an effective means used to define a set of candidate areas likely to contain faces, hands, or other human organ in a scene. This can be performed by using human skin color models or by threshold of the appropriate color space. A few papers comparing different approaches have been published, however, a study of color space influence on skin detection is still missing. In this paper we present two contributions the first one a comparative study of skin detection obtained by series of tests performed on 11 color spaces (YCbCr, HSV1, HSV2, RGB, RGBn, YUV, Ydbdr, Ypbpr, Hlab, YIQ, Yxy) chosen among the most used, using two methods: skin segmentation by skin Gaussian model created from a large variety of skin samples kept from images of different races people the second method is using threshold according Color space where we give our second contribution that consist to propose threshold for six color spaces, that we didn’t find in scientific literature. Experimental results reveal that using threshold or skin model, HLab is the most appropriate Color space to skin detection in color image. By our contribution we present a significant and useful tool to direct any faces detection system based on skin segmentation to use the appropriate color space that can increase positive detection rate and decrease the negative one.