Kernel Independent Component Analysis Based On Symmetrical

Face recognition is the most promising development of biometric identification technology. Because face recognition applications in national security, military and economic fields with a great application prospect, and involves many disciplines knowledge. Therefore, the study of face recognition technology has a theoretical significance and application value.In face recognition, differential extraction of effective face recognition feature directly affect the final recognition results. Accordingly, feature extraction has been one of the hot subject in this field, and this paper also focuses on this subject. The main work of this paper is as follows:1.According to face sample data present a nonlinear dividable characteristics, based on the existing subspace analysis method of thorough research, analysis the advantages and disadvantages of the existing algorithm, the existing independent component analysis method in the foundation, the union nuclear technology, nuclear method is introduced into the independent component analysis algorithm, and efficiently extract face nonlinear characteristics, so as to improve the algorithm recognition effect.2.Aiming at the problem that face recognition are generally lack of training samples, this paper analysis the geometric features of the face, and utilizes the face mirror symmetry to enlarges the number of training samples. In this algorithm, mirror transform is firstly introduced. Then, even/odd symmetrical samples are produced based on the theory of the even/odd decomposition principle, and the even/odd independent components are extracted from the corresponding samples respectively. So, it perfectly reflects the geometric structure of images,, and effectively reduce the impact caused by illumination variation and angle difference, so as to improve the algorithm of recognition rate.