摘要
针对传统工业产线上的质量检测环节,本文提出了一种基于FPGA控制板卡与物体识别算法的机器视觉检测系统。该系统采用FPGA+GPU并行处理结构作为硬件平台,充分利用FPGA实时性好、可实现复杂控制逻辑等特点完成视觉系统各部件的控制以及数据传输,配合电脑端的GPU完成物体识别算法。实验结果表明,该方案可以满足生产线上的高速要求,并且易于部署,从而实现了深度学习在传统生产线上的落地应用。
Aiming at the quality inspection of traditional industrial line,a machine vision detection system based on FPGA board and the object recognition algorithm is presented in this paper.The system is developed by the FPGA+GPU dual core hardware structure,taking the advantages of real-time data processing capacity and better logic control ability of FPGA,and the advantage of parallel computing and programmable capacity of GPU to implement the target detection algorithm.This system is then successfully deployed into a ceramic cup production line.Test results show that the proposed scheme can meet the requirements of easy-to-issue,high speed and good detection rate,demonstrating a potential application of deep learning in traditional industrial production.
作者
何其伟
赵宇坤
宗兆翔
HE Qi-wei;ZHAO Yu-kun;ZONG Zhao-xiang(School of Electronic Information,Shanghai Dianji University,Shanghai 201306)
出处
《数字技术与应用》
2020年第8期128-131,共4页
Digital Technology & Application
