The Face Recognition Algorithm Based On Wavelet Transform And Improved PCA And SVM

Nowadays, with the development of social and economic, the demand of information security is increasing for people and the development of the biometric recognition technology have achieved rapidly because their features are unique. The advantages of the face recognition technology among the number of the biological recognition technologies is non-contact, directly, friendly and convenient so it can be accepted easily by users and used to be identified widely in our lives. At present, with development of a variety of the face recognition technology, the performance of the face recognition can be accepted in our face recognition system in the controllable environment.Firstly, the images or the videos are collected from the face recognition system which should make some preprocessing, and then the face images should be detected and located if they exist. Secondly, the features of the face are extracted from these images. At last, the features can be classified and the information of the user can be identified.The various algorithms in the entire process of face recognition system are made an introduction and some research.The main works are following as:(1) When the images are collected, they are influenced by the location, light, noise and other objective factors. So the images need be preprocessed. The algorithms of the preprocessing of face recognition are described which include pixel gradation normalization, geometric normalization, images denoising and so on. The face detection algorithm based on adaboost is introduced in this paper and it is used to detect face in real-time. The accuracy rate is very high when detecting face.(2) The features extraction algorithm of combinations of the wavelet transform and the improved PCA is described. The basic principle of the PCA algorithm is described and an improved PCA algorithm is introduced in this paper. Because the low-frequency face information is obtained by the wavelet transform which can be useful for classification. When the PCA algorithm is used to extract feature, the dimension of the total scatter matrix of the samples is too large as a result rapidly declining in system performance. So the algorithm of combinations of the wavelet transform and the improved PCA is proposed for the feature extraction in this paper which can solve the problem. The experiments show that the recognition rate has not decreased and the time of feature extraction is shortened.(3) The basic principle of support vector machines is focused on. And how to use the support vector machine to solve the one-to-many problems on face classification. Because the support vector machine is good at solving the problem of small sample, nonlinear and high dimensional recognition, so the algorithm of combinations of the wavelet transform and the improved PCA is used to extract the face features, and the support vector machine is used as a classifier. The experiments show that the method of face recognition is proposed in the paper can obtain a high recognition rat on ORL face database, Yale face database and our face database.(4) A prototype system is implemented which include the face recognition algorithms are proposed in this paper. The designing scheme is given in this paper at last..