Face Recognition System And Key Technology Research

With the development of image processing technologies and the improvement of machinelearning algorithms, face recognition technology has been the concern of many fields, such as socialnetworking, security and e-commerce.In general, the key technologies of the face recognition system can be divided into two majorcategories of technology: face detection technology and face recognition technology. Face detectionrefers to judge whether a static image or a video frame contains a person’s face, as well as thelocation and size of the human face; face recognition refers to identify and determine the identity ofthe human face. In order to be able to improve both the accuracy of face recognition and recognitionspeed so that it can achieve the rapid real-time face recognition system,this article will study itfrom two angles of the key technologies of the face recognition system.First，this paper studies the current situation of face detection and face recognition technology,while analysis of the advantages and disadvantages of these methods. Combined with previousstudies mainly from the following three aspects of the key technologies in the face recognitionsystem:(1) In order to improve the accuracy of the face detection purposes, using adaboostalgorithm trained face optimal classifier towards measured image face detection, detected facecandidate rectangular area with adaboost algorithm trained human eye optimal classifier for thehuman eye to detect, further determine face candidate rectangle face really contain face objects,namely face detection algorithm based on Adaboost human eye to detect;(2) In order to achieve asmuch as possible to retain the critical areas of the face at the same time reduce the featuredimension, using the LBP histogram combined algorithm based on Adaboost algorithm face criticalorgan detection and unified mode facial feature extraction, after PCA algorithm featuredimensionality reduction, last been used to describe face information feature matrix;(3)Using thefeature classification recognition algorithms based on support vector machine match facecharacterized by the second step matrix with pre-save a good facial feature library in order to beable to confirm the identity of the face. Ultimately it constitute the core of this study facerecognition algorithm. According to a previous study of face recognition technology, and finally thispaper achieves an intelligent face recognition system.Intelligent recognition system achieved by this paper makes use of ORL database samples fortraining and testing. The test results are as follows: average recognition rate of0.332s, therecognition rate of93.13%. The results show that: the intelligent face recognition system based on face recognition algorithm proposed in this paper has a high recognition rate, at the same time hasgood robustness and recognition speed.