Research On Face Detection Algorithm Based On Improved BP Neural Network

After an overview on research results about face detection in our country and foreign countries, this thesis researches face detection based on BP neural network. Due to the defect that BP neural network easily falls into the local optimility, PSO algorithm is introduced to BP learning algorithm to improve the learning algorithm. Face detection system is achieved using the improved algorithm. The main contents of this thesis are as following:(1) The mechanism and basical structure of neural network are discussed. The application of neural network is introduced. And its learning algorithms especially BP learning algorithm are analyzed.(2) An face detection system for gray image is achieved using BP neural network as a tool. The problems how to choose face sample, preproduce sample, train neural network, and problems in the process of face detection, and their solutions are studied. The experiment results proved that the realized face detection system has better detection efficiency and robustness.(3) The traditional BP neural network is prone to fall to local optimility, to a certain extent, affecting the performance of the face detection system. Therefor Quantum Particle Swarm Optimization is introduced to traditional BP neural network learning algorithm to improve it. A face detection system is realized using the improved BP neural network learning algorithm, which achieves a certain Anticipated effect.(4) A new adaptive PSO algorithm whose performance is far superior to the traditional PSO algorithm is proposed. use of the optimization algorithm BP neural network was improved and implemented Face Detection System. The experimental results show that the system, even to the gray image with complex background, has fairly good results and good robustness.