Finding a good method for Skin Detection

Detection of skin in image is an important component of systems for detecting, and recognizing faces and hands. This work presents a study of three important methods of the color pixel classification approach to skin segmentation: Support Vector Machines, Multi Layer Perceptrons, and K-nearest neighborhood algorithm. Our analysis of two different types of representative color spaces using these classifiers shows that skin segmentation based on Support Vector Machines and RGB color pixel classification makes a real-time and accurate system.

Skin Detection, Neural Networks, Color space, Supervised Classification, Multi layer perceptron, K nearest neighborhood, Support Vector Machines, Correct Detection Rate, False Acceptance Rate, False Rejection Rate.