Design Of Face Detection And Recognition System Based On AdaBoost And SVM

Face detection and recognition technology has always been a hot research direction in the field of machine learning.In recent years,with the development of artificial intelligence industry,face detection and recognition technology has received widespread attention in society.The detection rate and speed of face detection directly affect the user's experience,so the implementation of fast and accurate face detection is of great significance.In this thesis,the face detection algorithm of AdaBoost and the face recognition algorithm of SVM are optimized,and implement the face detection and recognition system through the hardware platform of FPGA.Firstly,in the optimization design of face detection algorithm,image zooming is used to replace the traditional enlarged detection window,and fixed window scanning scheme is used to improve the detection speed.A method of the combination of rough detection and fine detection is adopted.After roughly detecting the face in the whole picture,eight detection windows around the roughly inspected face frame are detected accurately.To determine whether the face is detected or not,the false detection rate is reduced.Secondly,in the optimization design of face recognition algorithm,bilinear interpolation method is used to regulate the size of the input face.And the voting mechanism is used to classify all categories.Besides,dynamic sorting strategy is used to reduce the amount of computation of face recognition,and improves the recognition speed.Finally,ORL face database is used to train and test the face recognition system.Based on the Opal Kelly development board XEM6310-LX150,this thesis realizes the hardware circuit optimization design and system verification of the face detection and recognition algorithm of AdaBoost and SVM.The experimental results show that the face detection and recognition algorithm take up 4377 Slice resources(19%)and 3569 Kb RAM resources(74%)on Xilinx Sparten-6 XC6SLX150 type FPGA at 100 MHz.The detection rate and recognition rate of the face detection and recognition system for 1280 ? 720 images are 98% and 96% respectively.The detection and recognition speed is 14.6 frames per second.With the increase of the number of faces in video or image,the detection and recognition rate of the system will decrease,but both remain above 90%,and the detection and recognition speed will decrease slightly,but it will remain stable at about 14 frames per second.