RBF is the most widely used pattern recognition in a neural network, this article applies to such networks face recognition, face recognition model set up. Center for the hidden layer to improve selection algorithm, using Yale Face Image Database for the simulation experiments, analysis by comparing the recognition rate of their respective indicators to illustrate the method proposed in this paper does not affect the recognition rate in the case of face recognition can significantly improve the speed, reduce the system's storage capacity to meet the practical requirements of face recognition.