| With the proposal of intelligent manufacturing and the rapid development of artificial intelligence,the intelligent grasping system composed of traditional industrial robots and visual measurement technology is widely used in industrial manufacturing.Among auto parts,auto brake discs are made of cast iron,which is of high quality.Manual grasping is not only time-consuming and laborious,but replacing manual grasping with intelligent grasping can significantly improve production efficiency and save labor costs.In the intelligent grasping system,the visual measurement system replaces the human eye to identify the workpiece,and the visual recognition accuracy is the key to accurate grasping.As an active measurement method,structured light vision measurement has the advantages of non-contact,high precision,high resolution and low cost,and is widely used in biomedicine,machine vision,reverse engineering and other fields.In this regard,in order to restore the complete point cloud data of automobile brake discs,this paper proposes a method for obtaining point clouds of disc components based on structured light binocular vision.The specific work is as follows:(1)Research the principle of structured light binocular vision measurement,build a hardware system,and calibrate system parameters.According to the measurement principle,compared with the existing structured light coding methods,the phase shift coding with the advantages of high resolution and high precision is selected.Aiming at the important influence of phase unwrapping on the measurement accuracy of phase-shift fringe projection,a phase unwrapping algorithm based on complementary Gray-code is proposed after analyzing the time-consuming and insufficient unwrapping accuracy of the spatial phase unwrapping algorithm.The introduced complementary Gray-code is complementary to the fringe order of the traditional Gray-code,and the combination of order complementarity and median filtering can eliminate the phase unwrapping error caused by periodic misalignment;during phase unwrapping,the use of mask images to guide the unwrapping process can improve the Expand rapidity.Experiments show that the phase unwrapping speed with mask image guidance is improved by 36.1% while ensuring unwrapping accuracy.(2)In order to improve the matching accuracy and efficiency,a phase matching algorithm based on Gray-code is proposed.The horizontal constraint is formed by the principle of Epipolar geometry correction,and the vertical constraint is formed by the Gray-code fringe order,and the bidirectional constraint makes the matching area more accurate.Through experiments,the traditional AD-Census phase matching algorithm and the single-direction Epipolar geometry constrained phase matching algorithm are compared,which shows that the point cloud data of the vehicle brake disc obtained by the proposed algorithm is more complete.(3)In order to further verify the accuracy of point cloud acquisition,point cloud reconstruction was performed on standard geometric objects of known size,and their geometric parameters were calculated and compared with the actual parameters.The radius error of the hemispherical point cloud was 0.284 cm,and the open angle error of the conical point cloud was 0.094 It is proved that the method for obtaining the point cloud of structured light binocular vision proposed in this paper can accurately and effectively recover the point cloud of the object surface,which lays the foundation for intelligently grasping the brake disc of the car. |
