A skin detector based on neural network

A large body of human image processing techniques use skin detection as a first step for subsequent feature extraction. The objective of this work is to provide an efficient tool to detect human skin in color images. Well-established methods of color modeling, such as histograms and Gaussian mixture models have enabled the construction of suitably accurate skin detectors. However such techniques are not ideal for use in various environments. We describe a method of skin detection using a back propagation neural network, and show considerable good performance for a large variety of color images. We also introduce genetic algorithms into the weights and biases optimization of the neural network. The paper focuses on the novel approach to design a neural network based skin detector, which is later used to retrieve skin-like homogeneous regions in color face images.