Improvement of analyzing method for human skin color separation by independent component analysis

In this paper, we introduce the improvement of the conventional skin color separation method using two input skin images. The conventional method needs to satisfy the severe limitations that input skin image has no shading component and has plenty variation of the pigments density distribution simultaneously. In the two input images at the proposed method, the one has no shading component and the other has plenty variation of the pigments density distribution. To verify the effect of our method, we taken four subjects and separate these images to pigments components and compare the independent evaluation value (IEV) of separated signals extracted by using our method and the conventional method. The result shows that we obtain the separated signals which have lower IEV by using our method than the conventional method.