Structural Light Visual Sense And Image Processing Used In Crawling Arc Welding Robot

The crawling arc welding robot is to be used in all position, and it can move and work in the welding materials of the shape of upright wall and curved surface. Its duty is to deal with the automation of all-positon weld including the shape of the curved surface in the real-time welding job. The shapes of weld include the vertical, horizontal, and curved surface. This thesis uses the theories of digital image processing to research the robot's trace technology of welding seam, especially to the image processing of kinds of welding seam shapes.The thesis systematacially introduces the process of structural light visual sense and image processing used in crawling arc welding robot. The whole system begins at CCD's collecting images, and they will transmit into the image-collecting card. By this card's A/D conversion, the image can get standard format. After it, we use the software designed to process the image to get a clear one that will be recognized in next step. Finally we can calculate the parameters that the robot needs to control the operation.The working environment of the crawling arc welding robot is very bad. Strong arc disturbance almost cover the laser stripe, and the burned fog and splashed scrap iron both do harm to the image's quality. Besides it, the reflectance of laser stripe on the work piece, the non-linearity of the performance of camera lens and the random noise of in the process of photoelectricity conversion deteriorate it. For it I study kinds of the image processing of image enhancement and morphological methods, and deal with it with achievements. Different welding types and welding seam should be adopted by the different arithmatics, such as the butt, overlap, and abnormal V type, and so on. This thesis also studys it and makes lots of experiments.In the course of software development, we write it only by VC language, and avoid mixing other language's function into it. This makes operations more swift and convenient. The experiments show the software can adapt to real-time application.