针对自动焊接过程中焊缝特征点的检测问题, 提出了一套由工业相机和三线激光发生器组成的焊缝实时跟踪传感器设计方案。基于三角测量原理, 设计传感器的机械结构, 推导得出由激光条纹特征点像素坐标计算出焊缝特征点三维坐标的数学模型, 为有效抵抗在焊接过程中存在的弧光和飞溅的干扰, 采用高斯核相关(KCF)目标跟踪算法, 实时跟踪计算出图像中焊缝特征点的位置。实验结果表明, 采用最大焊接电流350 A的MIG焊, 在有强烈弧光和飞溅的工况下, 该装置能实现焊缝实时准确跟踪, 激光条纹与焊接熔池的最小距离可达20 mm, 传感器采样频率最高可达20 Hz, 跟踪误差小于0.42 mm, 能满足在实际焊接过程的使用要求。

Aiming at the problem of the detection of weld feature points in automatic welding process, this paper propose a real-time seam tracking designing scheme consisting of industrial camera and three line lasers. The mechanical structure of sensor is designed based on principle of triangulation. First, the mathematical model of 3D coordinates about seam feature points are calculated from pixel coordinates of feature points about line lasers tripe. Then, the Gaussian Kernelized Correlation Filters (KCF) is adopted to resist interference of arc light and splash, and real-time tracking compute the position of the weld feature points. The experimental results show that the device can realize real-time accurate seam tracking under the condition of strong arc light and splash, when the largest welding current is 350 A in MIG welding. Laser stripe and welding molten pool of minimum distance of 20 mm, the sensor sampling frequency can up to 20 Hz, and the tracking error is less than 0.42 mm, which can satisfy the requirement in the actual welding process.