Welding Seam Tracking System Based On Image Technique

Welding is one of important process in manufacturing. With the development modern technology, people pay more and more attention on welding quality and welding process. At present, welding atomization, welding robotization and welding intelligentizing are becoming the trend of welding industry. Vision sensing method is widely used because a great number of information can be acquired without contacting with workpiece. The paper consists of two parts: hardware part and software part. Hardware part is composed of image acquiring facilities (camera, image acquiring card) and LED light and robot. Software part makes dynamical control on robot by dada acquired by image information. To begin with, image data are acquired by image acquiring facilities, and then, original welding seam image characters are obtained by calculating acquired image data with image process program. Next, calibrating program turn image coordinate into world coordinate, which is consistent with robot environment. Finally, robot joint variables are acquired by robot anti-calculating. With the use of the variables, the system starts to track welding seam. The paper studied method of image processing applied to welding seam tracking, which mainly consist of image analysis and process, gray morphologic and pattern matching. Method of calibration applied for welding seam tracking is analyzed and studied in the paper. The conversion model is created among world coordinate, camera coordinate, image coordinate and pixel coordinate according to the principle of photic. Calibration and image process time are shortened with ROI technique in terms of great number of image information data. The system is of positive meaning for welding which man can't finish, such as nuclear environment or space environment. The system is of auto-modified function basing on sensing information, which provides essential study for entirely auto-welding of welding robot in the future.