Target Face Detection Using Pulse Coupled Neural Network and Skin Color Model

This paper presented a target face detection method combining pulse coupled neural network (PCNN) with skin color model. First we use "reference white" method to achieve light compensation, and reuse a mixture skin color model based on HSV and YCbCr to detect skin color areas, and then use mathematical morphology and a face detection algorithm to obtain the sub-graph of face. Finally, we use PCNN to extract time series of benchmark face and all sub-graphs, and achieve the target face detection through computing the correlation of time series. Simulation results show that this method has higher accuracy independent of the benchmark face training set, and reflect better detection performance.