Abstract: A comprehensive analysis on the Hough transform for detection of straight line in an image is presented. Firstly, the corresponding three relations between the parameters of straight line in an image field and in the transform field were summed up. By those, the aimless computations and searches in Hough transform processing can be reduced greatly and its efficiency is improved. Secondly, in the design of accumulator, a triplet data structure is used and can reduce much amount of accumulator. Thirdly, a new detection method of welding seam is proposed based on these results, by the method, the offset of welding seam is not only detected, but its direction is also obtained. In the implementation of the method, the objective gray image is transformed black-and-white image with the edge detection of Prewitt and median filtering algorithm, therefore the welding seam is enhanced and that the noisy is depressed. Finally, simulating and practical application results show that the detection precision of the method is higher and the anti-inte- rference ability is good.

Key words: Edge detection, Hough transform, Line function, Welding seam, Wobble plate Hydraulic joint Load matching efficiency Position tracking