Application of SOM network to study facial skin segmentation after facial surgery

The importance of facial surgery is clear to everyone these days. So there is a need to accurate and proper color facial skin segmentation to be able to extract variety of features from it. Many facial skin segmentation methods have been provided till now published. This paper suggests a method to segment the skin accurately using a new efficient color space and SOM network in two phases. The proposed method is robust in detecting the accurate bounds of facial skin in frontal and lateral color images. The 96.85% percent accuracy of the proposed method on the experimented data indicates its efficiency comparing to considered methods in this paper including SOM standard method and MLP network.