Skin Segmentation based on Gaussian Model by Combining Color Spaces

This paper proposes a novel idea of combining the color spaces as a pair in getting the improved results in the human skin detection. Skin detection methods that can be used in various applications of image segmentation are evaluated. The performance of the three pixel based classification algorithms are analyzed. The color images of human faces are segmented into skin regions on the basis of piecewise linear decision boundary, Gaussian model to detect human skin regions, and finally, the concept of combinations of two color images are used to configure and analyze the effect of them on the segmentation of human skin regions. The extensive set of experiments is performed over the commonly used face color images using color spaces, namely YCbCr, YUV, YES, YIQ, and XYZ. The color spaces are combined together in using the Gaussian Color Model. The experimental results shows that the method of combining more than one color space with the Gaussian model produces better results than the ordinary skin cluster and Gaussian model with highest skin detection rates of 100%.