Research On Real-time Visual Tracking System Based On FPGA

Visual tracking is not only an important research directions in the field of computer vision,but also the core of the field of the video surveillance technology topics. In recently years,science and technology are advancing by leaps and bounds,visual tracking technology hasalso been considerable development, and accompanied by the development of videosurveillance technology, have higher requirements of real-time monitoring system for theimage processing speed. In this case, the programmable logic device capable of parallelcomputing has unparalleled performance and advantages, more and more concern andattention by the parties. On this basis, this paper choose Xilinx Inc. XC4VSX25series FPGAchip as the core processor, designed a visual tracking system, compared with the traditionalARM, DSP and PC platforms, the design is more flexible, faster, easy to upgrade, readilyavailablechange the design content.This paper includes the following aspects:(1) Build FPGA-based hardware platform for visual tracking system.Start of acquisition fromthe video data stream, until the final display, the intermediate relates to the variousmodules of the architecture, including an acquisition, the display, and the frame memory.These efforts is the premise of the dectction and tracking algorithm.(2) Video decoder chips are configurationed by I2C bus controller that simulationed by FPGA,change PAL format video signal into standard YCbCr4:2:2format digital signals.(3) Design DDR SDRAM controller. The completion of the calculation and storage of thebackground frame parameters.(4) Moving object detection algorithms in FPGA hardware platform.(5) Use Kalman filter to predict the position of moving objects and parameters in the nextframe of binarizing grayscale images.(6) The function simulation of each module, and testing of each data use Chipscope Software.