Research On Plane Welding Path Automatic Recognition Based On Image Processing

The welding technology is playing a significant role in metal heat processing. Arrange commodities to highly sophisticated products, the welding technology is indispensable. Meanwhile blended with other technologies such as artificial intelligence, computer vision, digital information processing and robotics, the welding technology is stepping for high efficient in processing, high intelligent in quality control and robotization in manufacturing. Although the robot has powerful functions, it is limited to use widely due to specific program, running cost and investment cost etc.Aiming at improving the welding efficiency and quality and extending the application fields,in this thesis the images of welding path were obtained by non-contact CCD digital imaging technique and were recognized.The welding path recognition system was developed with Delphi 7.0 in the environment of windows2000, which functions include filtering, sharpening, binarization and edge detecting. The straight line and curve can be identified through Hough transform and line approximation method and the results are saved for NC auto programming.This system can be used in other plane welding when working with the corresponding jig and rational welding craft arrangement, and this makes numerical welding machine becomes more efficient and flexible by changing NC codes according to different welding path and resolves the contradiction between process automation and process cost in some extent. Facing the serious energy problems, constructing economical society is a necessary trend. As far as increasing availability of resources and saving energy are concerned the thesis has significant meaning and will certainly bring the considerable economic benefits and social benefit.