Research Of Face Recognition System Based On Video

With the need of video surveillance and information security, the research of face recognition in video cause More and more people's attention. Many excellent face recognition system have been developed at home and abroad, And get a wide range of applications. The purpose of this study is to develop a system, to track and recognize the face in video, the main tasks are as follows:(1)On the basis of former analysis, this paper summarize the classic algorithm about face detection,tracking and recognition technology. And it give the advantages and disadvantages of each algorithm and the applied direction;(2)In the aspect of face detection and tracking,this paper used the algorithm based on AdaBoost to detect the face in video, and determined its size and location. Based on in-depth analysis of MeanShift tracking algorithm,it used its improved CamShift tracking algorithm. And combied CamShift algorithm with the algorithm based on AdaBoost, it can track the face in video very well.(3)In the aspect of feature extraction, based on vector of the principal component analysis (PCA), two-dimensional principal component analysis (2DPCA)and the two direction two-dimensional principal component analysis ((2D)2PCA) based on image matrix are introduced. And for different feature vectors have different contribution rate to face recognition, propose a two-dimensional principal component analysis (W(2D)2PCA).Then combined image sub-block with local features fusion technology, adopt fuzzy classification decision rules ,get good recognition results in the contrast test on the ORL face database.(4)In the developed environment based on VC++6.0, supporting with OpenCV library, it developed a face recognition system in video to track and recognize 30 individuals in the lab. The experiment proved that the system has a good real-time and accuracy in tracking and recognition in video.