Research Of Occluded Face Detection And Recognition Based On Video

As one of the most promising biological characteristics identification technologies, theface authentication technology has developed at full speed in the last few decades. It hasthe important academic research value and broad application prospects. In practicalapplications, the face authentication system collects the face images in uncooperativeconditions where face images are easily covered, which leads to the incompletion ofgathered face data and accordingly affects the whole system’s accuracy rate. In view of thelocal occluded human face, this paper focuses on the detection and identification of theoccluded face based on video. The main research achievements include:(ⅰ) The rapid occluded face detection based on multi-feature Adaboost. Using the threeframe substraction by the information of the video sequences, the algorithm estimates thegeneral location of the face. For the long time of the adaboost characteristics training andthe occlusion issue, this paper gives an improved adaboost algorithm to confirm theaccurate position of the face. The experimental results prove that, when the face isoccluded, the method proposed in this paper is significantly better than the traditionalAdaboost algorithm.(ⅱ) The occluded face recognition based on MB-LBP characteristics. According to thelow recognition rate of the occluded face, this paper gives an algorithm based on MB-LBPcharacteristics. Finally, this paper gives a classification based on orthogonal projection toreduce the number of feature matching. The experimental results prove that, when the faceis occluded, the method proposed in this paper can improve the recognition rate effectively.(ⅲ) Realized a face authentication system based on video. This paper gives theframework and detailed design of the system based on the proposed detection andrecognition algorithm, establishes the facial sample library, and lays the foundation forfurther study on the research of face authentication.