摘要
针对焊接过程中弧光、飞溅等干扰导致焊缝特征点提取困难的问题,提出了一种基于激光线段方向聚类与核相关滤波器的焊缝特征点定位的方法。该方法在初始帧中对图像中激光条纹分割,再用线段方向聚类分割出不同线段确定初始帧中焊缝特征点的位置;针对焊接时的强干扰,使用两个核相关滤波器实时跟踪焊缝特征点,最后用卡尔曼滤波器融合两个跟踪器的输出结果,得到焊缝特征点的位置。实验结果表明,该方法能在坡口焊、搭接、T型等常见的焊接接头中准确找到焊缝特征点位置,跟踪算法在强弧光干扰下的焊缝特征点跟踪精度较高,误差在8个像素以内,满足焊接需求。
To solve the problem of difficult extraction of weld feature points caused by arc and splash interference in welding process,a method of locating weld feature points based on laser line segment direction clustering and kernal correlation filter was proposed.In this method,the laser fringe is segmented in the initial frame,and then different segments are segmented by line segment direction clustering to determine the location of weld feature points.In view of the strong interference during welding,two nuclear correlation filters were used to track the weld feature points in real time.Finally,the output results of the two trackers were fused with kalman filter to obtain the position of weld feature points.The experimental results show that this method can accurately find the location of weld feature points in common welding joints such as groove welding,lap welding and T-shaped welding,andthe tracking algorithm has high tracking accuracy under strong arc interference,and the error is within 8 pixels,which meets the welding requirements.
作者
杨智
於双飞
李坚
刘楚天
YANG zhi;YU Shuang-fei;LI Jian;LIU Chu-tian(School of Electro-mechanical Engineering,Guangdong University of Technology,Guangzhou 510006,China)
出处
《组合机床与自动化加工技术》
北大核心
2021年第11期35-38,共4页
Modular Machine Tool & Automatic Manufacturing Technique
关键词
焊缝跟踪
线段聚类
核相关滤波器
卡尔曼滤波
weld tracking
line segment clustering
kernal correlation filter
kalman filter
