Skin Color Detection by Combining a Statistical Skin Color Model, Neural Network Technique and Moving Object Information

In processes of face detection, face tracking and gesture recognition, the skin color detection is usually a post-processing step that plays such an important role. The skin color detection can almost be viewed as a two class problem with a non-skin color pixel vs. skin color pixel classifier. In this paper, we present a classifier that was built by combining a statistical skin color model and neural network technique. Moreover, we also combined this classifier with moving object information, which was extracted from video sequence data. This combining approach has showed a better testing classifying result in comparison with the others that use only one of the above techniques. Our approach can be implemented on DSP platform to make a fast smart vision sensor.