A study on the modified Hough algorithm for image processing in weld seam tracking

摘要

Nowadays, automatic Gas metal arc (GMA) welding is receiving a growing attention for its fast production speed and wide-ranging versatility. Furthermore, automatic and robotic welding systems are urgently needed for general application, which depends on the realization of auto-weld seam tracking. Seam tracking system allows for not only robot or machine trajectory shifts, but also adaptive control to change weld bead formation, making further improvement on the weld quality. Based on the laser vision system, this study aims for development of an image processing by the modified Hough algorithm which applied for seam tracking system in GMA welding. To achieve this objective, image features in the procedure of welding processing were discussed and specially treated. Algorithms in the image processing were investigated to maximize the processing effect based on the modified Hough algorithm, which was quite costefficient in line detection. Quantitative evaluations of the used algorithm were set up to acquire optimal image processing algorithm. Finally, a common image processing method was employed to verify efficiency of proposed image processing and extract the welding location in a shorten time for industrial application. It can be concluded the detected feature points and centerlines proved that this algorithm can be used for automatic GMA welding process as well as industrial applications.