Study Of Arc Welding Robot Welding Seam Detection Technology Based On Vision

In recent years,with the rapid development of welding automation,image processing andsensor technology, welding products manufacturing automation,intelligence and informationhas become an inevitable trend. Owing to the features of mass information, high reliability andbroad applicability in the arc welding robot weld inspection, the vision sensor-based weld seamdetection has been widely applied and gradually become a hot topic.It is important to explorehow to detect and identify weld information accurately, quickly and efficiently, and improve thequalityofcenterlineautomatic,which aretheresearchpriorities.In this paper, we describe the status of weld detection and identification of domestic andinternational and the view of related experts in the field. Based on VC++6.0of the softwaredevelopment platform, we design the real-time welds image processing system.The system usesCDD sensor to capture the weld image,process the A/D conversion by the image acquisitioncard, then the digital image obtained is transmited into the computer system, and it will processtheimagebytheimageoperation methodandget theweldseaminformation.Image processing and feature extraction is a core of this paper, and we achieve to processthe image base on VC++6.0. Firstly, in the part of image de-noising filter and the median filter,the appropriate method is applied to filter image by comparison of the advantages anddisadvantages of domain average filtering and median filtering. Secondly,We obtain the binaryimage by adjusting the threshold manually. And in the part of image edge detection, we use themethods of the Sobel operator, Robert operator, Prewitt operator, and LoG operator to detect theimage edge. Based on edge extraction renderings, we choose the LoG operator image edgedetection method. After edge detection, we process the image by the method of contourextraction, image fining, and edge fitting calculations to ensure the accuracy of the algorithm.Finally, it will achieve the desired purpose of the weld image detection, extracting the weldedgesandtheweldcenterlinesuccessfully. In terms of programming ideas, this paper modulizes the functional module by object-oriented analysis method and packages the module into different classes. It separates logic intoachieve by base on-oriented interface programming. It is only interested in the interfacebetween the modules, rather than concerns inside of module. And the method is conducive tothemaintenanceandexpansion ofsystem.This design of the weld image detection system can detect weld image effectively, andextract the edge and the center line of the weld successfully. After verify the accuracy of image,the algorithmns can achieve the purpose of detecting weld information reliability. If it can beapplied to the welding robot control system, the torch perform welding tasks by the weldcenterlineidentified,it willpromotetheaccuracy andautomation ofweldinggreatly.