Research On Three-dimensional Detection And Recognition Methods Of Three-dimensional Parts Assembly Qualification Rate

The three-dimensional detection technology of industrial parts has always been the hotspot of research and research frontier.The three-dimensional detection and recognition method of the components and parts assembly rate includes two part,one is obtaining the three-dimensional reconstruction point cloud.For industrial product detecting,it is in particular need for high-precision,fast three-dimensional reconstruction method.Second is the assembly defect detection method.This dissertation has been carefully studied for these two parts.Firstly,according to current exists three-dimensional reconstruction technologies,three-dimensional reconstruction based on multi-frequency heterodyne method is selected for the point cloud acquisition under the two evaluation indexes:precision and speed of detection.At the same time,aiming at the problem that the three-dimensional reconstruction speed is slow in multi-frequency Heterodyne method,proposed a multi-frequency heterodyne algorithm based on heterogeneous platform,and the phase unwrapping,phase superposition,phase unwrapping and other processes of heterodyne multi-frequency are transplanted to the GPU,and the operation speed of the phase diagram extraction process is improved by the characteristics of single instruction and multi data flow of the GPU platform.At the same time,in the process of image point matching,multithread parallel technology is introduced,which greatly improves the running speed of the algorithm.Secondly,for the defection part,choose two kinds of defect to study,one can be identified by two-dimensional image defects recognition algorithm,such as screw sliding teeth and so on.The other kind of defect cannot be directly identified by two-dimensional image recognition algorithm,such as workpiece depth detection and so on.For the first case,the two-dimensional image recognition algorithm is used to identify,and the mapping relationship between the three-dimensional coordinate system and the image coordinate system is mapped to the point cloud space to calculate the defect degree and the installation qualification rate.For the second case,the point cloud is first segmented,the background cloud is removed,and then the point cloud recognition algorithm is used to identify the defect and the defect is calculated for the defective workpiece.Finally,the screw strip defect detection and pilot lamp defect detection as examples,according to the three-dimensional defect detection algorithm proposed to build detection system,verify the validity of the proposed algorithm.The test results show that the three-dimensional detection and recognition method of the qualification rate of the stereoscopic parts assembly can well identify defects and calculate the degree of defects,and is robust,achieving the detection target and detection task.