Moving Target Detection And Tracking System Based On FPGA

With the development of computer vision technology and increasing demand ofintelligent, the intelligent video surveillance system is widely used in production and life.Moving target detection and tracking system as the key point of the system has receivewidespread attention. Now, the moving target detection and tracking related algorithm hasgradually matured, but realizing the algorithm by hardware is still a focus and difficulty inthe research.This thesis research on FPGA moving target detection and tracking system based onFPGA. The system, which selects Zynq-7000series chip of Xilinx company, makes full useof the characteristics of FPGA parallel processing and Xilinx advanced development toolsSystem Generator. Detection algorithm based on three frame difference method andtracking algorithm based on template matching successfully transplanted to FPGA.Design of moving target detection and tracking system can be mainly divided into twoparts: the algorithm design and hardware realization. The algorithm design: First,this thesisdetermined using three frame difference method and template matching method to realizethe moving target detection and tracking based on the analysis and comparison.. Secondly,in the process of object detection, using the Gauss filter for target detection preprocessing tosolve the noise problem exists in image acquisition and morphological processing for targetdetection postprocess is used to solve the phenomenon of cavitation is easy to appear indifference method. Finally, template matching method be used to achieve the goal.Hardware realization: The hardware platform consists of ZYBO, MT9V034CMOS cameraand VGAdisplay. Among them, the ZYBO evaluation board realize moving target detectionand tracking. It mainly divided into six parts: Bayer Gauss RGB module, filter module, amotion detection module, morphological processing module, motion tracking module and VGAdisplay module.In this paper, ISE, System Generator, MATLAB and other tools be used to completetiming simulation and comprehensive verification of each module on the ZYBO evaluationboard. Simulation and experiments show that, the system implementation effect comparedto the traditional frame difference method and template matching method has higheraccuracy, better robustness to noise.Moving target detection and tracking system designed in this paper based on FPGArealized the detection and tracking of moving target. The system has strong robustness,real-time. It can be used in the field of intelligent video monitoring.