Skin Extraction Based on Fuzzy Cluster

Skin color is an important feature for face detection in color images. Usually by applying a statistical skin model, possible skin region can be segmented. However, the output can not be as accurate as expected. Besides, simple 2D model, e.g. CbCr, can not present the real skin point distribution. So this paper presents a 3D CrCbCg model to describe skin distribution more precisely. Meanwhile, considering skin points in a specific image have a relative stable distribution, a cluster-based skin model is presented to remove background points which are wrongly retained by the general model. Because of applying the characteristic of each specific image, this algorithm can effectively improve the performance and accuracy of skin model. Experimental result shows that this algorithm can achieve satisfied results for over 95% images, including obviously improved images for over 70% .