Research And Application Of Feature Extraction In Face Recognition

The automatic recognition of human face is one of the most attractive and challenging problems in the fields of pattern recognition and computer vision . The aim of many scientists who work in computer science field is to make robots have human's intelligence and recognize and remember person just like what human does . In the authentication, access control systems, detection of criminal investigation, video surveillance and other areas,face recognition technology have a very wide range of applications. The research on face recognition, its significance lies not only in promoting the image processing and pattern recognition theory and applications to meet the authentication and so on, such as content-based retrieval of the actual demand, at the same time as a result of the specificity of face patterns, the application of face recognition study on the promotion of physiology, cognitive science, psychology and other related disciplines also have a positive impact.With the improvement of science and technology and development of society .It is urgently needed for convenient and reliable automatic-status face recognition methods .So face recognition resumes is to be the highlight of machine intelligence research.In this paper,on the basis of the reviewed and discussion of face recognition and feature extraction that existed brought a new method on purpose , and proved it through experiments. Summing up the main work is as follows:The first section of the paper is the summary of the full text . It is the introduction and analysis the development and the status of the classification of the face recognition technology and feature extraction technology.The second section describes a number of statistics-based feature extraction and selection methods, including principal component analysis (PCA), Linear Discriminate Analysis (LDA), maximum distance criteria (MMC) and so on.The third section is the analysis of the human face image feature extraction ,in the part of human face image feature extraction, this paper get a new method of feature extraction on Maximum Margin Criterion with Locality Preserving .Compared with original MMC method, by multiplying the defined weights and regulating the parameter, the new method here can still better manifold local structure information.The fourth section is the experiments analysis of the new method that got from the third section. The experimental results on Olivetti Research Laboratory (ORL) face database ,YALE database and UMIST data base show that the new method is robust to illumination and pose ,and can recognise the face images efficiently and enhance the recognition rate.