Research On Edge Detection Technology Of Welding Pool Image Based On Mathematical Morphology

With the development of machine vision and image processing technology,it has become an important research direction to achieve closed-loop control of welding quality by directly observing the welding pool with CCD camera through edge detection and other technical analysis.Due to the external environment during welding and spark splashing during the welding process,the edge detection effect of the molten pool image is not good.Therefore,it is particularly important to find an algorithm which can detect the edge of the molten pool image accurately.In this dissertation,the weld pool image is taken as the target,and the traditional edge detection algorithm is used to detect the edge of the molten pool image.Aiming at the poor detection result of the molten pool image edge,a detection method that can preserve the edge of the molten pool image is proposed.The edge detection algorithms of traditional and mathematical morphology are used to perform edge detection on the molten pool image in this paper.In order to solve the problems such as unclear detection results,breakpoints and noise,binarization is used in the edge detection process.In view of the serious loss of information on the left and right sides of the molten pool image in the traditional binarization process,an improved approach is put forward.The partial adaptive method is used for binarization,then median filtering,establishing connected areas and searching for padding to obtain the final binarization result.The binary image was detected by using the edge detection operator combined with four kinds of morphological basic operations and selecting the appropriate shape and size of the structure elements.Simulation and evaluation results show that the proposed approach is feasible,the edge information of the molten pool well is kept and extract the more accurate weld pool edge is extracted,which lays the foundation for the final welding quality control.