Processing And Identifying Image Of Welding Molten Pool Based On Morphology

The research on arc-welding automatic control system based on CCD vision sensor has become one of the hot spots in the field of welding both at home and abroad in recent years. Extracting borderline of the welding molten-pool image tracked by CCD is the first step of welding process control and quality control, so more and more scientific researcher and technical staff began to pay close attention to the algorithm design and steps. Concerning the shortcomings of current algorithm in that it can not handle filtering noise while extracting borderline exactly, three aspects are carefully studied in this paper:1. Based on the analysis of the characteristics of welding molten pool, the different pretreat smooth algorithms and binarization algorithms are programmed to demonstrate the respective results in the molten pool image, and prove that the median filtering algorithm is more advantageous in filtering point noise.2. Based on the analysis of shortcomings of traditional gradient operator in its application in digital image processing, the non-linear digital image processing algorithms are carried out by designing program: Binary Morphology and Gray scale Morphology. Thus we prove that the gray scale morphology gradient operator can keep the original shape of the molten pool and filter the noise point at same time while extracting the borderline of the welding molten pool.3. Based on the analysis of actual demand of welding molten pool image processing system platform, the design principle and funcional requirement of the image processing platform are brought forward. In the end, the object oriented programming languages C++ is used to realize the welding molten pool image processing system platform in Windows operating environment.