The Optimization Algorithm Of Face Recognition Based On Gabor Wavelets And SVM

With the rapid development of economy and the popularity of the Internet,information security has gradually been valued by people. As one of the important means toprotect the safety of information, face recognition have gradually been valued byresearchers. Face recognition is an important branch of computer vision technology andbiometric identification technology which is also is an important field of pattern recognitionand artificial intelligence.The important task of the face recognition is to identify the facesin the static image or video. How to recognize face quickly and correctly is a difficultproblem in face recognition task at present. The selection of face recognition algorithm isdirectly related to the recognition rate.First of all, This paper introduces the development status of face recognition in and outabroad. the paper details the conventional method of pretreatment. Method of face imagepreprocessing are commonly used: gray level transformation、 binarization、 imagenormalizing、image filtering、image sharpening，etc.Secondly, principal component analysis (PCA) algorithm, two-dimensional principalcomponent analysis (2DPCA) algorithm, linear discriminant analysis (LDA) algorithm isstudied in this paper. The three algorithms are fused, forming a "PCA—LDA" and"2DPCA—LDA" algorithm. Three experiments were conducted to verify the performanceof these types of algorithms.Finally, In this paper, Gabor wavelet and support vector machine (SVM) werestudied,Gabor wavelet has good function of biological neurons and has good adaptive ofchanges, SVM has good classification effect, This paper combines Gabor wavelet and SVM with PCA algorithm and2DPCA algorithm, which is named Gabor+PCA+SVMalgorithm and Gabor+2DPCA+SVM algorithm. Through identification of the ORL facedatabase, Yale face database, self built database, verify the effectiveness of the algorithm, atlast,build a face recognition system based on the establishment of a Gabor wavelet andSVM.