武汉工程职业技术学院学报2026，Vol.38Issue(2)：39-44,6.
基于机器视觉的焊缝跟踪系统设计与实现
Design and Implementation of a Weld Seam Tracking System Based on Machine Vision
摘要
Abstract
With the rapid advancement of industrial automation,welding robots are increasingly widely used.However,weld seam deviations caused by workpiece assembly errors and thermal deformation seriously compromise welding quality.A machine vision-based weld seam tracking system is developed to address this is-sue.The EFORT ER6-1400 welding robot is adopted as the execution platform,and the laser structured light vision sensing scheme is applied.For image processing,a complete preprocessing procedure consisting of median filtering for denoising,image enhancement based on the improved dark channel prior,binarization via maximum-median filtering and Canny edge detection is designed.Automatic recognition of four types of weld seams,namely butt,corner,lap and groove seams,is realized by means of the ResNet-50 network.The pixel-to-robot base coordinate mapping is established via camera calibration and hand-eye calibration,and a PID trajec-tory correction algorithm with feedforward compensation is designed.Experiments show that the system achieves a mean tracking deviation below 0.3 mm and a response time of approximately 35 ms,with notably improved weld formation quality over conventional teach-playback methods.关键词
焊缝跟踪/机器视觉/图像处理/轨迹修正/焊接机器人Key words
seam tracking/machine vision/image processing/trajectory correction/welding robot分类
信息技术与安全科学引用本文复制引用
李业刚..基于机器视觉的焊缝跟踪系统设计与实现[J].武汉工程职业技术学院学报,2026,38(2):39-44,6.基金项目
安徽省教育厅科学研究项目"基于机器视觉的焊接系统应用研究"(项目编号:2025AHGXZK30819) （项目编号:2025AHGXZK30819）
