| As the development of technology proceeds, the embedded system of object tracking is more frequently applied in both civilian and military realm. A large number of international and domestic academics have been researching the object tracking system with great effort making it a very popular research direction in machine vision.This dissertation is devoted in designing an object tracking embedded vision system, which can be used in SD analog videos. After a comparison of many solutions for framework, DSP and FPGA was employed on the board. And since EP3C16Q240C8N and TMS320DM6437 were more mature, these two chips are considered the most viable choice for the design.During this project, the board’s schematic and PCB layout were designed. A complete test and verification for each module of the board was conducted subsequently. This essay focuses on the signal integrity analysis planning the PCB layout of the board along with the high-speed routing of DDR2 which is the most complex and vital part of the design.In addition to the hardware design, software design of embedded system was also an indispensable part. This section is focused on the bootload of the DM6437, DSP/BIOS, PSPDRIVER, FPGA’s driver and the interface design of DSP and FPGA. Beyond it, framework design of application program, which is based on RF5, was also completed along with the video input and output tests, making this a complete versatile system.At last, the tracking algorithm based on layered template matching, which has been transplanted and optimized on DM6437, was introduced with the optimization methods. Through actual testing of our board, hardware and software meet the reliability and stability performance criteria. With preliminary goal reached, the next step is to update the hardware of the whole system and optimize the algorithm. |