查看文章

The goals of this paper are to recognition the butt welding joint for welding robot environments by using a new approach of background subtraction seam path process. Butt welding joint images were captured from CCD camera mounted on the top of the work bench then processed by the proposed approach method to extract and determine the weld seam path position in xy coordinates. The image is segmented used Sobel filtering to determine amplitude images in access channel then subtract between the access channel and input images with different threshold values to find the region of edges joint. Next process is to apply morphological technique, fill up the hole to calculate the area, skeleton and generate the sub-pixels type of data (XLD) contour points. The position of the start, mid, end and auxiliary points of butt welding joint are selected according to contour points with the three case studies identification weld seams path process. The results, shows that the proposed method was capable to automatically detect and identify the butt welding joints in three case studies without any prior knowledge of the shapes. In terms of match error compared with the actual position, case study 2 had the lowest matching error which is less then±2 pixels either column or row. The highest match errors happened in case study 3 where the matching error is in the range of±11 pixels.