Design & Implementation Of Dynamic Image Measurement And Processing System Based On Zynq

Early vision measurement systems were mainly implemented by PCs.It was not suitable for complex scenes,had poor real-time performance,and was not easy to carry.In view of the above shortcomings,embedded vision measurement systems based on ARM,DSP,and FPGA have been valued and developed rapidly.Xilinx's Zynq(FPGA + ARM)embedded programmable on-chip processing platform can execute complex control programs,dynamically schedule computing resources,and support algorithm hardware acceleration.Compared with the use of ARM,DSP,FPGA and other embedded processing systems,Zynq has the advantages of good scalability,high performance,low power consumption and low cost.Therefore,this paper designs and implements a dynamic image parameter measurement system based on Zynq.This paper first analyzes the mathematical model of binocular stereo vision measurement,and designs the overall scheme of the image parameter measurement system based on Zynq and the functions of each module by combining the software and hardware working principles.To overcome the shortcomings of common stereo image matching algorithms for binocular vision measurement,a stereo matching algorithm based on multi-feature fusion is proposed.This algorithm uses the Census transform,the edges of the image,and the HSV color space to form a matching cost calculation module for multi-feature fusion.Compared with the algorithm that uses a single image feature to calculate the matching cost,the algorithm in this paper improves the matching accuracy.The proposed algorithm was packaged into a hardware-accelerated IP core using Xilinx's Vivado HLS tool.Design and Implementation of Software and Hardware Cooperative Work for Image Parameter Measurement System Based on Zynq.Implementation based on Zynq binocular acquisition system;then complete the system's binocular camera calibration;Design and implementation of image preprocessing hardware accelerated IP core;According to the common Zynq image processing architecture,the image measurement system of this paper is designed and implemented,and the distance parameter measurement between the target object and the camera is completed.The system test is performed on the dynamic image parameter measurement system based on Zynq.The experimental test is performed based on the experimental platform and experimental scene.The measurement and analysis are performed for different relative distances.The measurement range is 400mm-900 mm.The relative measurement error is less than 2.5%.The rate is 18 fps,which meets the real-time measurement requirements.Finally,the design performance indicators of the system were compared to meet the design requirements.