| In the process of industrial welding,3D weld identification is one of the important contents.With its large projection area and fast reconstruction time,the 3D reconstruction technology of weld based on surface structured light has been paid more and more attention in the 3D positioning of weld.In this paper,aiming at the problems of low accuracy and instability in the extraction of plate workpiece splicing welds,a 3D visual recognition and location method of plate workpiece splicing Welds Based on face structured light is proposed.This paper mainly studies:(1)By studying the model of surface structured light projection measurement system and system calibration technology,the development of surface structured light 3D reconstruction system is designed and completed.In this paper,the principle of surface structured light 3D reconstruction and the key technologies of camera calibration and system calibration are described in detail;The center phase of sub-pixel points is fitted by bilinear interpolation.Finally,the system parameters are fitted based on SVD algorithm.On the basis of ensuring the accuracy of calibration parameters,the influence of sub-pixel calculation phase error is avoided,and a three-dimensional measurement platform based on area structured light projector,camera and computer is successfully built.(2)By studying the image processing of two-dimensional weld image,a weld extraction method based on depth map is designed and implemented.In order to solve the problems such as the two-dimensional feature is not obvious and the stability is poor,the two-dimensional position information of the start and end points of the weld is fitted by using the filtering algorithm,the combination of edge detection and depth gradient detection.(3)By studying the relationship between the three-dimensional model of robot coordinate system and the transformation of coordinate system,the hand eye calibration method and coordinate system transformation method based on 3D vision are designed and implemented.Based on the robot motion and the detection of the calibration plate,the hand eye coordinate transformation is realized.Aiming at the problem that a single field of view cannot be displayed completely due to the large workpiece size,the coordinate transformation and fusion of point clouds in multiple fields of view are carried out,and the unified coordinate visualization of point cloud model is realized.(4)A three-dimensional measurement system composed of a structured light projector and a large constant image camera is designed and implemented.At the same time,the control software of the system is designed.The software development is completed by taking the experimental simulation of large-size plate magnetic pole workpiece as the measurement object.It is mainly composed of image acquisition module,system calibration module,point cloud generation module,weld identification module and other basic functional modules.It can generate the overall point cloud of the workpiece in real time,and the point cloud accuracy is about 0.33mm.The overall system realizes the complete workflow of the 3D visual recognition and positioning system of the plate workpiece splicing weld based on the surface structured light.In this paper,the precise positioning of the magnetic pole splicing weld is realized.A set of start and end point data of 18 groups of weld measurement on the simulated magnetic pole workpiece is collected experimentally,and the welding is carried out by robot.The extraction of welding points is stable and the welding effect is good. |
PDF Full Text Request