Software And Hardware Co-Design Of Fatigue Driving Detection System Based On Zynq

As the number of cars continues to grow,the number of traffic accidents has increased year by year.Traffic accidents pose a major threat to people's lives and property safety,and research shows that fatigue driving is one of the main causes of traffic accidents.Therefore,it is of great significance to design a fatigue driving detection system with low power consumption,good real-time performance and portablity.The ZedBoard Development Board is selected in this thesis,its main chip is xc7z020-clg484-1 of the Zynq-7000 series,and the structure of the Zynq chip is ARM+FPGA.ARM is well-controlled and easy to bulid an operating system,and FPGA has a powerful parallel computing capability.The fatigue driving detection system is implemented by a software and hardware co-design method,which could give full play to the advantages of ARM and FPGA.FPGA mainly includes face detection IP,Video Direct Memory Access(VDMA)and HDMI interface.VDMA is a mass data transfer channel between FPGA and DDR,and HDMI interface is used to complete the high-definition display.Face detection IP is designed by Vivado High-Level Synthesis(HLS)tool,which uses the method of image scaling to implement the AdaBoost algorithm.Face detection IP mainly includes image scaling,integral image calculation,image traversal,classifier detection,window merging.The image scaling adopts bilinear interpolation method.The integral image can be quickly calculated by the integral image increment algorithm.Setting the row and column step is to accelerate the traversal speed of detection window.Window Merging uses the mean method.Aiming at the hardware design,classifier detection adopts the structure of serial and parallel combination to accelerate the detection of pending window.ARM mainly includes Linux operating system,fatigue feature judgment,hardware IP driver and operating interface.Hardware IP driver controls VDMA and face detection IP operation,and the operating interface is the user and the whole system communication interface.Fatigue feature judgment mainly includes face segmentation,illumination correction,edge detection,feature fitting and judgment,fatigue judgment.Face segmentation uses the law of "three court five Eyes" to divide the human eye and mouth area.Illumination correction adapts an adaptive Gamma correction algorthm.Edge detection using Canny algorithm.Eyes and mouth feature extraction use ellipse fitting and circumscribed rectangle respectively.The ratio of the long and short axes and the height of the rectangle are used to judge the state of the eyes and mouth.The result of feature fitting judgment combined with PERCLOS criterion to judge fatigue state.In this thesis,the fatigue driving detection system is realized by the method of software and hardware co-design on the ZedBoard Development Board,the system has portability,low power consumption and real-time.