Face Recognition Using PCA And Neural Network Ensemble Based On DCT

Face recognition is one of the most challenging problems in the fields of pattern recognition, image processing, and computer vision and cognizance science. It has turned into an active research topic and developed quickly in the recent decade along with various applications. But, to set up practical automatic face recognition system, there are still many problems unsettled, especially the efficiency and robustness of the algorithms.The technology of face recognition analyzes face images with computer, and then extracts effective recognition information from them, which is used to recognize the identification of a person. The research of the face recognition technique mainly focused on how to extract the features describing face image. It needs that these features are tolerant of geometry changes, impression changes and illumination changes; meanwhile the features should contain the information that distinguishes this face pattern from others. This paper presents several approaches usually applied in face recognition system: geometrical feature based, eigenface based, elastic matching based, neural networks based approaches, and analyzes their advantages and disadvantages. Based on algebraic features of the images, this paper first introduced the PCA-Based face recognition algorithm. The computing time of the traditional PCA-Based face recognition is very large, so the discrete cosine transform is used to do the preprocessing. The method proposed in this paper deals with the original face images using DCT first, and then applies PCA to a few DCT coefficients that contain most of the information of the original images. The experiments results show that the correct recognition rate of the proposed method is superior to that of methods using DCT only and almost equal to that of methods using PCA only while the speed of the training is one time faster than the traditional PCA algorithm introduced in the precious chapter.The design of classifier is crucial to face recognition. The BP neural network is widely used as classifier which has self-study, robusticity, self-adaptability. In this paper, we replace individual BP which prone to local minimum with BP ensemble. Experiments show that the recognition rate of the BP ensemble is 3 percent higher than individual BP while the recognition time is notably shortened.This paper adopts Matlab language to realize face recognition arithmetic, and the simulation experiments are conducted based on face images in ORL face database. The results show that the recognition rate is quite high and the train time is notably shortened, so the method is efficient for face recognition.