The Research And Implement Of Face Recognition Based On Video Stream In Classroom

As the society step into the twenty-first century, with the fast development in computer hardware and software technology, the requirements on public security, information security and privacy protection are increasing day by day. As one typical technology of biometric identification technology, face recognition gets more and more attention by researchers. As one of the most host area of computer vision and pattern recognition, it has a great potential application market.This paper will mainly apply this technology to the real-time video stream analyze. The aim is to build a real-time face recognition system in the classroom environment. The system can effectively increase the interaction between teachers and students, which we think will improve the teaching quality effectively. The system based on the real-time video stream analyze, first do face detection by Adboost algorithm, then using LRPCA to extract the face feature from the gray-scale face image. After all these, compute the Miscellaneous-distance between the test sample and the samples in face database to come the result of face recognition. During the step of LRPCA feature extract, the image will also be preprocessed including normalized, gray-histogram equalization, declining face rotate by ASEF algorithm. The paper will also analyze the performance of different feature extract algorithms.The paper will bring up a real-time video stream oriented face recognition system which can be used in normal classroom. In the system, the face recognition algorithm is designed in independent module, so it is very easy to change it with other recognition algorithm to do the compare and data analyze.