计算机工程与应用 ›› 2019, Vol. 55 ›› Issue (12): 208-214.DOI: 10.3778/j.issn.1002-8331.1804-0003

• 图形图像处理 • 上一篇    下一篇

机器视觉中边缘检测算法的SDSoC加速实现

吴  进,赵  隽,李  聪,吴汉宁   

  1. 西安邮电大学 电子工程学院,西安 710121
  • 出版日期:2019-06-15 发布日期:2019-06-13

Implementation of SDSoC Acceleration Algorithm for Edge Detection Algorithm in Machine Vision

WU Jin, ZHAO Jun, LI Cong, WU Hanning   

  1. School of Electronic Engineering, Xi’an University of Posts and Telecommunications, Xi’an 710121, China
  • Online:2019-06-15 Published:2019-06-13

摘要: 针对机器视觉图像处理中边缘检测算法要求越来越高的实时性,提出使用SDSoC加速实现机器视觉中的边缘检测算法。基于SDSoC开发环境,选用ZC706作为开发平台对Canny边缘检测和Sobel边缘检测进行加速。SDSoC环境支持处理器系统(Processing System,PS)和可编程逻辑(Programmable Logic,PL),根据PS和PL的特性,将两种边缘检测算法中的模块分配在各自适用的硬件架构上运行,即在PS端使用优化的数据分配方法,在PL端使用缓冲区结构及优化指令。实验结果表明,对于512×512的图像,Canny算法用时4.61 ms,Sobel算法用时3.20 ms,满足了机器视觉算法实时性的要求。

关键词: Sobel边缘检测, Canny边缘检测, SDSoC

Abstract: In order to improve the real-time performance of edge detection algorithms in machine vision image processing, SDSoC is proposed to speed up the implementation of edge detection algorithms in machine vision. Based on the SDSoC development environment, the paper uses ZC706 as a development platform to speed up Canny edge detection and Sobel edge detection. The SDSoC environment supports Processing System(PS) and Programmable Logic(PL). Based on the characteristics of the PS and the PL, the two edge detection algorithm modules are allocated to run on their respective hardware architectures. Use the optimized data distribution method on the PS side and use the buffer structure and optimization instructions at the PL side. The experimental results show that for 512×512 images, Canny algorithm takes 4.61 ms and Sobel algorithm takes 3.20 ms, which satisfies the real-time requirements of machine vision algorithms.

Key words: Sobel edge detection, Canny edge detection, SDSoC