Research On 3D Reconstruction System Of Welding Seam Based On Laser Scanning

As a common machining method,welding is an indispensable part in industry.It plays a very important role in material forming and is widely used in many fields.The quality of the welding and the polishing after welding have a direct impact on the performance of the object.However,most of the current inspection and polishing are manually performed,and the degree of automation is not high.The quality of inspection and polishing is directly determined by the experience and proficiency of the workers.In addition,the poor environment also affects the efficiency and health of the workers.Since most of the information obtained by human comes from vision,and because machine vision technology is becoming more and more mature,the application of machine vision technology in weld inspection and grinding can be closer to manual operation.Compared with the traditional two-dimensional vision,three-dimensional vision has more depth information and can reflect the shape of the object more truly.The 3d reconstruction of the weld can be used for both the welding quality inspection and the model input of the automatic grinding system.Therefore,based on the principle of laser triangulation imaging,this paper studies the 3d reconstruction system of welding seams by laser scanning.The research work of this paper is as follows:(1)The robot and two-dimensional laser sensor are used to build a point cloud acquisition platform to achieve multi-perspective scanning of objects to obtain point cloud data.By calibrating the hand-eye system composed of the robot and the laser sensor,the coordinate system of the laser sensor is converted to the base system of the robot,so that the robot can recognize the position of the welding seam..(2)Point cloud registration is studied.Firstly,the essence of point cloud registration is analyzed by mathematical model,and then a point cloud registration method based on multiscale point characteristics is proposed to solve the problem of low computing efficiency of existing registration methods.This method is characterized by high computational efficiency and improves the overall registration efficiency while maintaining the registration accuracy.(3)According to the characteristics of weld point cloud,the greedy projection triangulation method was used to reconstruct the point cloud surface.In order to solve the problems of low efficiency and rough surface in reconstruction of greedy projection triangulation,a combination of bounding box downsampling and Moving Least Squares(MLS)was used to improve it.Experiments show that the improved algorithm can improve the reconstruction efficiency and ensure the quality of the reconstructed surface.(4)The three-dimensional reconstruction system of welding seams was designed,and relevant algorithms were verified by using the actual collected weld point cloud data.