Color-based skin detection: survey and evaluation

This paper presents a comprehensive review and evaluation for color-based skin detection. The literature is reviewed in four categories: 1) colorspace selection, 2) skin color modeling, 3) dynamic modeling and 4) illumination invariance and adaptation. Using a large data set of 1894 images, we examine whether the colorspace transformation can increase the compactness of skin class and the discriminability between skin and nonskin classes in fourteen 3D colorspaces and fourteen 2D chrominance planes. Classification performance of SPM, GMM, SOM and SVM is also evaluated. The results reveal that 1) colorspace transformation cannot improve performance in general, 2) the absence of the luminance component decreases performance significantly, 3) the performance variance in 2D chrominance plane is more greater than in 3D colorspace, 4) except for 3D Bayes SPM, the color "preference" of detector is intrinsic and quite similar, and 5) the performance of SPM is superior to that of others