Research Of Video Face Recognition Based On Weighted Voting And Key-Frame Extraction

As one of the most potential biometric technology, face recognition receives extensive research and application in recently years, especially face recognition based on video. This paper mainly researches the face recognition technologies based on video at a short range. My research achievements include:(â…°) Face key-frame extraction algorithm based on color histogram. Through researching on the video key-frame extraction algorithm, this paper puts forward the face sequence key-frame extraction algorithm based on color histogram and applies the algorithm to the face recognition system. By key-frame detection and extraction, the algorithm removes the unfavorable frames for face recognition in the face sequences, reduces the amount of calculation and improves the recognition efficiency.(â…±) Video face recognition method based on weighted voting. Through researching on the effects of feature points in the different parts of face region, the number of feature points and Gabor feature selection bring to the face recognition results, this paper improves the traditional elastic graph match algorithm, and puts forward a new video-based face recognition method based on weighted voting. The experiment results show that the improved algorithm not only increased the recognition rate, while reduced the time of recognition.(â…²) The implementation of video-based face recognition system. By comprehensive utilization of the algorithms introduced in the previous chapters, combined with DirectShow and OpenCV development technologies, this paper designs and realizes the framework of video face recognition system based on weighted voting and key frame extraction which lays the foundation for further research.