Image processing and weld deviation recognition of robotic deep penetration TIG welding

摘要

With the development of machine vision, vision-based robotic welding technology is widely used in manufacturing industry. In this study, weld deviation recognition of a new type of deep penetration TIG welding was studied. Firstly, a bilateral filter was used to remove the noises and preserve the edge of the region of interest (ROI) of images obtained by a high dynamic range (HDR) CCD. Secondly, we used the improved Otsu algorithm to obtain accurate welding arc shape, weld pool and weld characteristics. Then Canny algorithm was applied to obtain a complete edge contour. Finally, the arc center line and the pixel coordinates of the midpoint of the weld are obtained using parabolic fitting and maximum curvature algorithm, respectively. The experimental results show that the proposed algorithm can accurately obtain the deviation of the arc center line and the welding seam coordinates, which can meet the accuracy requirements of robotic welding. In order to realize the automatic tracking, the distance between the center line of the arc and the middle point of the welding seam is obtained, and the real-time correction function of the welding robot is realized.