Face Recognition System Based On Gabor Wavelet And Non-linear Algorithm

Face recognition is an identification technology which analyzes face image with computer, and then getting effective recognition information. With the development of some applications such as electronic commerce, face recognition has become one of the most potential biometrics authentication methods. it can be applied in many aspects, such as criminal recognition of police system, monitor system of bank and custom, automatic doorman system and family entertainment, etc.The face recognition system includes two parts: face detection and face recognition. Face detection is the first and key step of face recognition system, the precision of face detection algorithm directly affects the following image's disposal and analysis steps. This paper adopts the face detection algorithm based on AdaBoost which is effectively for prophase work of the face recognition system. The importance of face recognition is feature extraction. In this paper, we analysis many of face recognition algorithms, and then according to the result of experiments, we present a new face recognition algorithm based on Gaobor wavelet and Supervised Local Linear Embedding,this algorithm's effectiveness has been verified by experiments. This paper's primary work as following:1) Analysis and realize the face detection algorithm based on AdaBoost. And then the experiments on the pictures from internet show the advantage of this algorithm.2) Introduce the definition of Gabor Transform and the method of feature extraction by using Gabor wavelet transform. The Gabor wavelet transform has two advantages: Firstly, it can describe the receptive field of simple cells in the visual cortex of human cerebra correctly; Secondly, Gabor wavelets shows the joint time-frequency property of signal analysis and overcomes the shortcomings of traditional Fourier Transform while fails to present any time discrimination ability in frequency domain. Analysis coefficients value of 2D Gabor wavelet and its application in face recognition, emphatically.3) Introduce and analysis many classic face recognition algorithms, such as PCA,LDA,KPCA,KLDA,LLE, etc. According to the experiments conduct on the Matlab platform, sorting out the advantage and remaining problem of each algorithm. In order to solve the problem of the sample information not being fully used and deficiencies of other algorithm which be presented before, a face recognition algorithm base on Gabor wavelet and SLLE(Supervised Locally Linear Embedding) is realized, the algorithm includes two steps: first of all, getting Gabor feature by Gabor wavelet transform to face image and down-sampling of Gabor feature, and then reducing the features'dimensions again by SLLE. The result of experiment on the ORL and YALE face database shows a 3.5 %~37.8% increase in recognition rate, compared to others, by using the proposed algorithm, which improves face recognition performance effectively.4) After analyzing experiment reusult, we finished the realization of attendance system based on face recognition. The recognition rates are above 90% through series experiment on database of human faces which includes 20 persons and 10 pictures for each one, also time expense of each experiment is limited in 1 second, all the above show relatively high recognition rates and real-time performance.