The Gait Recognition Algorithm Based On Support Vector Machine

With the development of society,people's information security requirements are getting higher and higher,biological recognition technology has always been the concern of researchers at home and abroad.Gait recognition is a technology that identify mainly rely on people walking posture,it have many advantages,such as long-range identification ?non-contact and low pixel requirements,so it have great application value in reality.we start to research algorithm for gait recognition based on support vector machine(SVM).This article designs and builds a video image acquisition pretreatment circuit system with FPGA development board and CMOS camera.Through the QUARTUS II platform and the Verilog HDL language to accomplish the pretreatment algorithm for video stream,which includes image acquisition,image processing,data storage and results display these functional modules.The image processing module completes the extraction of the human body contour by using the background difference,mean filtering and Sobel edge detection algorithm;completes the human skeleton model extraction by using the region filling and refinement algorithm that shoot by CMOS camera.The human body skeleton model video stream through the VGA interface connect the display to accomplish real-time display,and the human skeleton model video stream stored in the SD memory.The data of the human skeleton model video stream that stored in the SD memory is used to study the gait feature extraction algorithm by MATLAB.Firstly,the program based on the contour sharpness algorithm is designed to get the coordinates of the joints of the human body and the proportion.Secondly,the body characteristic angle algorithm program is designed,Through the hip joint coordinates to find the height of the foot joint.Finally,Combined with the eigenvalue of unified HU moment,Select the radial basis function and theleft one calibration method of the support vector machine to complete the automatic recognition of the gait system.After the actual image acquisition and processing experiment of the system,it is verified that the combination of feature angle and unified HU moment eigenvalue can effectively recognize the gait by support vector machine(SVM),which provides a reference for the study of gait recognition.