The Research And Design Of Face Detection System Based On FPGA

Automatic face detection and recognition system, because of its convenience, high-speed, safety and reliability is being more and more widely used in the area of identity authentication, access control, real-time monitoring, virtual reality, human-computer interaction, remote video confere-nce, image database retrieval and so on, which also shows a very broad space for development and application prospect. In almost all of face information processing cannot do without this important aspect of human face detection, therefore face detection has become an independent study and research in the area of face information processing.This paper, starting from the real-time and reliability of human face detection, maily focus on the research and realization of face detection algorithm and FPGA system design, the main work includes:(1) Three kinds of skin color segmentation algorithm based on YCbCr space are analyzed and compared. For the problem of misjudge in the high and low brightness regions generated by the elliptical skin model, this paper proposed a improved idea by choosing the value of long and short elliptical half shaft according to the brightness value. MATLAB simulation experiments show that this improved algorithm has higher accuracy of segmentation while the speed of segmentation is almost unchanged.(2) A new face detection algorithm is proposed by combining image background elimination with elliptical skin color segmentation and geometric feature distinguishing in YCbCr space.(3) According to the proposed algorithm, the design of system architecture and hardware logic design based on FPGA are completed.(4) The designed system is tested and verified actually in the Altera’s DE2experimental development board. Experiments show that this designed system can realize the real-time face detection in video images while keeping a high detection rate and low false detecting rate.