RECOGNITION OF NO ARC V-SHAPE WELD SEAM UNDERWATER

He Yinshui,Zhang Hua

DOI: https://doi.org/10.3969/j.issn.1001-1382.2006.02.009

2006-01-01

Abstract:A vision-based underwater weld tracking system was designed, which was mainly made up of laser, a piece of filter and CCD camera sealed in the waterproof glass container. The laser was used as the subsidiary light source, which irradiated the weldment as well as there is an angle of 30° between the laser and CCD to get the V-shape cross-section weld seam. Because the images shotted underwater were vague, median filter method was adopted to strengthen the images. After that,threshold transform was used to get the weld seam and then Krisch was used to get the edges of the weld seam and the center lines of the weld seam by averaging the gray number of the image edges. At last, a technology was used to make the gray points that were not on the same lines to get the characteristic points,that were the lowest points of the center lines on the weld seam as well as a line was drew to identify them.