Based On Face Recognition System Video Sequence

The face recognition technology is one of the important research directions in the field of pattern recognition and machine vision. In the ideal case (the same light, still pictures, the same attitude, etc.) face recognition technology has been very mature, but in the actual environment applications is difficult to widely used especially in the real-time video surveillance. Due to the plasticity of the human face and a variety of factors affect the imaging process, the face recognition system under the conditions of the real-time video, the recognition performance will be significantly reduced. So that the face recognition system is difficult to apply in the industrial, commercial and family life. A complete face recognition system includes a face image preprocessing, face detection, facial feature extraction, feature matching. The content of this study by the following aspects:(1) In this paper, the face image preprocessing methods:mean filtering, adaptive median filter, histogram equalization method to achieve the face image de-noising, image enhancement, image restoration. The experiments show that these pre-processing can greatly improve face recognition rate and save storage space. Through these methods, we can get the best face image from the video.(2) The face recognition feature extraction method, and these methods in detail, and simply pointed out the advantages and disadvantages of these methods in research applications, but also describes in detail the face recognition rate of performance indicators and these indicators do a detailed comparison.(3) Based on Adaboost face detection algorithm to detect the face in the video sequence, the use of Gabor filters for feature extraction of face and experiments to select the appropriate characteristics to complete the dimensionality reduction, face recognition rate, and in other methods (nearest neighbor and support vector machine) for comparison.(4) In Visual Studio2008development environment, assisted development source OpenCV library, development of face recognition system based on video sequences, and video recognition experiment face database. The experiments show that the system has good real-time and accuracy.