Research And Design Of Face Detection System Based On FPGA

In this paper,FPGA technology is combined with video image technology to design a real-time video acquisition and face detection system based on FPGA.Video image acquisition technology is the premise of image analysis,which mainly involves the problem of video clarity and fluency,as well as the degree of system integration,considering the cost and volume of the system.Face detection technology is a key link of automatic face recognition systems,in the face of a biological recognition technology,such as target tracking technology to lay a solid foundation,can also be applied in a particular security environment,such as entrance guard and traffic,such as face detection and tag for the crowd area information,in this paper,the main work is as follows:1.A real-time video image acquisition system based on FPGA was studied.Cyclone IV E series chip was selected as the main control platform,and an efficient and convenient real-time image acquisition system was built by combining DDR2 memory,VGA control chip and external crystal vibration.The design of each module driver circuit and driver program,to obtain high definition,high frame number of real-time video image.2.Study the implementation of image preprocessing technology in FPGA technology,including color space conversion and image filtering.The color image graying is involved,and the adaptive filtering technology is proposed in this paper in combination with the common filtering technology to make the image better smooth and true,and the contour information is obvious.Preprocessing the acquired real-time video image data not only makes the display vision more real,but also provides a good pixel basis for further development.3.Research and put forward the new face detection technology and the implementation of FPGA technology,mainly from the background interference and face recognition aspects.It mainly involves color space conversion,which separates brightness from color,color component for color recognition and binarization,background difference algorithm,binarization image and operation,and closed operation for expansion and corrosion to reduce background interference.The specific conditions are used for face qualification judgment,the detection rate is improved,and finally the face area is determined through the video image and the box is selected.This is the key technology of face recognition and target tracking.4.System debugging and experimental analysis were carried out.Combined with theexperimental results,it can be seen that the frame rate of the video acquisition system is 30 frames,the resolution is 1280*720,and it can well detect the moving face and select the box,which fully meets the design requirements of the system and has a wide range of application prospects and development space.Finally,the system design is summarized and the possibility of technical improvement is prospected.