Face Recognition Algorithm Based On Local Binary Pattern

Everyone has its unique biological characteristics, such as inborn features: fingerprints, face, DNA, etc; and acquired features: the signature, language, gait, etc. These biological characteristics have the non-duplicated uniqueness and stability which human inherent, it is not possible to copy, theft or forgotten, and we have made certain success in recognition of research. And in various biological characteristics identification method, face recognition has its simple, accurate, economic and expansibility advantages, so it occupies an important position in biometrics. After years of development, the face recognition technology has made considerable progress, and has made the good effect on entry control, safety inspection, crime search and other areas of application. The performance of face recognition method depends heavily on the extraction of characteristics, at present, there are many face feature extraction methods such as: PCA, ICA, LDA, LBP etc, with LBP feature extraction is the most important method of feature extraction.With reading a lot of domestic and foreign relevant literatures, this paper discussed the face recognition based on LBP method. First, it introduced the basic principle and developing evolution, and summarized the basic idea of LBP face recognition's algorithm, and then, elaborated several face identification algorithms improved based on LBP and analyzes the advantages and disadvantages of LBP face identification algorithms, Finally, this paper proposed an improved LBP face recognition algorithm, through sample pretreatment, including photo size of normalization, color image gray-scale aging and gray image histogram equalization processing, to improve and enhance the LBP feature space. Demonstrated by authoritative test on ORL and Yale face library, this method has better ability of face feature representing capabilities and distinctiveness.