The Study Of A Real-time Face Recognition System Based On IP Camera And SRC Algorithm

Face recognition has been viewed as one of the hot topics in the research fields of biometrics identification technology. As a non-contact certification, it is highly praised by users and researchers. Thus, many algorithms in face recognition were presented every year. In recent years, John Wright used sparse representation theory, which regarded face recognition as solving linear equations. Based on this, he proposed Sparse Representation Classification(SRC) which had a good performance under occlusion. In this paper, we exploit a face recognition system which uses sparse representation to perform classification. Generally speaking, the main work is just as follow:(1) We make research on the SRC algorithm. The SRC algorithm, which has robustness to the illumination and occlusion, can find the best item associated with the testing sample from the base. At the same time, we select Principal Component Analysis(PCA) and Local Binary Pattern(LBP) as the feature, Support Vector Machine(SVM) and SRC are used as classifier to study their effect in face recognition. Experimental results show that the SRC+PCA has the best recognition rate than the other. Because of real-time requirement of system, too much complex calculation can make the system slow, thus in this system we select SRC+LBP.(2) We present the row-overlapping blocks SRC algorithm and apply it to the face recognition under occlusion. The main idea is that, for all training samples and a testing sample, we devide them into some row-overlapping blocks in the same way. And then, we can obtain the results of each overlapping block’s class by SRC algorithm. Finally, we can identify the testing sample by voting from the results of all testing row-overlapping blocks. Experimental results on AR database show that the recognition rate of row-overlapping blocks is higher than the SRC algorithm and the weighted SRC algorithm. Further more, the average recognition rate is higher than the uniform blocks SRC algorithm.(3) We make research on IP camera and rewrite some functions of Software Development Kit(SDK). In this paper, we use IP camera as the acquisition equipment instead of USB camera. By making research on the IP camera, we get the frame and change it into the format that OpenCV can support.(4) We design a real-time face recognition system based on IP camera and SRC algorithm by way of OpenCV and C++ programming development. First, AdaBoost algorithm is used to detect face in each frame, and then LBP is used to extract the features of texture. Finally, we obtain the results by SRC algorithm. Experimental results show that the system can deal with real-time video and has robustness to the illumination.