Human Face Recognition Based On Two-Stage Feature Extraction And PSO-BP Neural Network

The pattern recognition is a new subject that developed rapidly for near 30 year. Face Recognition is one of the key issues in pattern recognition and artificial intelligence fields. It involves image processing, psychology, physiology, recognition science and so on, which has a wide range of potential applications in the areas of identification of certificate, security control and criminal capture.The main objective of our study is to propose an improved face recognition method based on the combination of Principal Component Analysis and Neural Networks. In order to enhance the extracted speed of eigenvectors, a two-stage kernel feature extraction methods for face recognition is developed in this paper. The algorithm includes two stages: firstly, the principal component analysis (PCA) is employed to condense the dimension of image vector. What follows, Kernel principal component analysis (KPCA) are applied to the reduced dimensional training samples. Finally, the experimental results on ORL face databases indicate that the proposed method is more efficient while retaining the same recognition.Based on tow-stage kernel feature extraction method, a face recognition framework is proposed. After the features of the images are extracted, an improved back propagation (BP) algorithm is introduced to train the neural network for recognition. The weights of neural networks are optimized using Particle Swarm Optimization (PSO) algorithm. The framework combines the optimization of the PCA+KPCA and the adaptability of the PSO-BP neural network to improve the recognition rate and the robustness of the algorithm to noises. Comparing with the traditional PCA and KPCA algorithm, the method proposed here can significantly have a better effect on extracting representation of the differences among different faces, and comparing with the traditional BP neural network, PSO-BP neural network also have the faster convergent speed and the higher prediction accuracy, which results in a high recognition and a high robustness of the algorithm. Experimental results presented in this thesis verified that the proposed algorithm is accurate and effective.