Face Recognition Based On Gabor Feature Global Weighted Representation Of Sparse Representation

With the progress of technology and development of the society, identity authentication in people’s contact has become the core contents that building the credibility, face recognition technology emerges as the times require, feature extraction as the critical step has become a research hot spot. Difficulty in face recognition feature extraction, at home and abroad, there have been many algorithms, the study of these algorithms, this paper, based on sparse representation can be global and local feature samples into account, in order to enable better identification. This paper studies the basis of face recognition and feature extraction Gabor wavelets on the basis of the original algorithm is proposed to improve the algorithm. Feature extraction problem solving research and for the development and maturation of the field of pattern recognition is significant.Thesis began to introduce the current development of face recognition, face recognition and describes the features extracted important influence on the development of science and technology and society, and on the recognition of several typical algorithms are analyzed and discussed; paper began introduced the development of face recognition, face recognition and describes the features extracted important influence on the development of science and technology and society, and on the recognition of several typical algorithms are analyzed and discussed; secondly discussed based Gabor feature extraction of the basic theory and algorithms ideas and feature extraction on the problems of the research priorities; and finally on the basis of existing theories and algorithms, we propose three improved Gabor features face recognition algorithms.First, take on local binary pattern(LBP) operator, principal component analysis(PCA) and partial Preserving Projections(LPP) extracted feature vector dimensionality reduction experimental analysis. Thereby enhancing the effectiveness and identify the time dimension reduction.Secondly, we propose a face recognition algorithm(GSRC) global Gabor features based on sparse representation, GSRC for Gabor feature sparse representation classification smallest l1 norm sparse solver accuracy problem. Firstly, Gabor wavelet transform processing face images obtained Gabor features, the establishment of over-complete dictionary, and then introducing the vector of the total variation model in the global features, and integration of Gabor features and global features, and finally the use of sparse representation model to optimize the characteristics of fusion. Experiments were carried out in the GSRC algorithm different face database experimental comparison can be drawn through the experiment, this new face recognition algorithm GSRC both for image illumination or pose and expression changes and other factors have a strong robustness.Third, on the basis of non-linear discriminant analysis algorithm is proposed kernel Fisher discriminant analysis Gabor face recognition algorithm combining mean, kernel Fisher Fisher linear analysis is performed in the feature space by this method not only has the ability to describe the nonlinear capabilities of face images, but also has categories based on Fisher linear discriminant analysis can be points of advantage.The experimental results show the feasibility and effectiveness of these three algorithms.