Image Processing And Identification For Welding Seam Based On Visual Sensing

In industrial production the existing welding robots,as a whole mostly are the first generation of playback robot or partly are second generation of off-line programming robot.These two types of welding robots do not have the ability to adapt to the harsh environment and operating conditions changes.Therefore,it is very important to use the laser vision sensing technology to realize the automatic welding process of the welding robot.In this paper,the paper focuses on the automatic tracking system of plate welding,focusing on the welding seam image processing and post-weld morphological feature detection in welding seam automatic tracking system.The main contents and results of this thesis are as follows:(1)The significance of research on automatic seam tracking is put forward,and the research status of vision sensing technology,image processing technology and 3D image reconstruction technology at home and abroad are introduced.(2)The overall structure of automatic seam tracking system is introduced.The image acquisition experiment platform of laser vision sensor system is set up,and the image acquisition scheme and system component selection are determined according to the requirements.(3)The pretreatment and post-processing of the welded structure light image are carried out.The preprocessing includes filtering denoising and image binarization.After comparing and improving with other filtering denoising algorithms,the adaptive median filter is used to process the weld image,and then the image is binarized by Otsu method.The post-processing includes edge extraction and centerline extraction of the laser bead.After comparison and analysis,the centerline extraction algorithm based on morphological skeletonization is used to process the laser stripe image.(4)Based on the extracted coordinates of the center line of the laser stripe,the Delaunay triangulation algorithm and the B-spline interpolation algorithm were used to reconstruct the target weld respectively.The three-dimensional effect of the welding seam was compared and analyzed,The B-spline interpolation reconstruction algorithm is significantly better than the Delaunay triangulation algorithm.(5)Based on the characteristics of weld defects in this paper,six kinds of evaluation indexes of weld appearance were established.The algorithm was used to extract the three-dimensional welds based on the topography evaluation index.The index values extracted by the algorithm were the results show that the data obtained by the two methods are basically the same,the biggest error is 0.041 and 0.035 respectively,which will not affect the results of the weld appearance evaluation and the follow-up Weld inspection.The result shows that the system can be used to measure the parameters of weld formation and to judge and predict the subsequent multi-layer and multi-channel weld parameters.