The Research Of Face Recognition Technology Based On PCA-SIFT Algorithm

Face recognition is currently a hot topic in the field of artificial intelligence, involving image processing, machine vision, pattern recognition and other fields of study. It belongs to biometric identification technology, by using the biological characteristics of individual organisms to distinguish different organisms (usually refers to person).It integrate the computer image processing technology and biostatistics, by using computer image processing technology to extract feature points of human face from the video, and using biostatistics to construct the mathematical model, namely the construction of facial feature templates. According to the result of comparison of the known facial features template and the facial characteristic of the measured object, get a similar value to determine whether they are the same person. With the rapid development of artificial intelligence, has been widely used in video surveillance, access control systems, criminal identification, screen session, file management, and photo detection and other fields, has a profound theoretical value and broad application prospects. Face Recognition technology research, including face detection, face normalization, feature extraction, feature matching and other aspects. But in cases of illumination, facial expression changed greatly, or there is some obscure or image blurring, the accuracy rate will be significantly reduced, and the problem of computation time of massive data. All these need to be solved urgently, so the main work of this paper is as follows:1) The subject using PCA-SIFT algorithm to characterize the human face. It is the improved algorithm of SIFT algorithm. In the premise of ensuring accuracy, using principal component analysis (PCA) to reduce the dimensions, changing the descriptor generation methods, in order to reduce the computational time, improve the algorithm in real time.2) Propose an optimized matching strategy. The initial matching based on the Distance-ratio criteria, then used RANSAC algorithm to exclude outliers, in order to reduce the false match rate, also save the time of matching and improve the accuracy.3) This article constructed a face recognition system based on the above algorithm. The Preliminary experimental results verify the proposed related technologies. At the end of this paper, a summary of the full text and the suggestion of the future work are provided.