Research On Algorithm Of 2D Face Recognition

Human face recognition is one of the most reliable biometric recognition technologies. It has become an important research topic all over the world. An automatic face recognition system consists of four sections: face detection, preprocess of face image, face feature extraction and recognition. In this paper, we focused on the face feature extraction and recognition.Foundational theories of Kernel Principal Component Analysis(KPCA) and Support Vector Machine(SVM) are firstly introduced. KPCA uses the thought of kernel function to make the data dimension-increasing to extract features in the feature space. KPCA can use the higher order information of data effectively and has much more advantages than pricipal component analysis. SVM is a novel learning method based on statistical learning theory. It has perfect theory foundation and excellent learning ability, which has been a hot topic in machine learning area. The combination of these two theories provides face recognition strong theoretic support.The kernel principal component feature extraction from the primitive face image and residual space face image has a strong complementarity. Therefore, a serial integration method for feature parameters is used in this paper. and A new select algorithm of integration parameters is proposed to look for the optimal integration parameters automatically. The two different kernel principal component features which extract from the primitive face image and residual space face image respectively are integrated by the optimal integration parameters. Thus an optimalion feature for face recognition is obtained. Finally, then face recognition is finished by support vector machine classifier. The experimental show that the proposed method can obtain high optimal recognition rate and average recognition rate and is effective.