Research On The Detection Method Of Workpiece Assembly Defects Based On Coded Structured Light

As an important part of intelligent manufacturing,3-D detection technology of workpiece assembly defects has always been a hot spot and frontier of domestic and foreign scholars.Its research contents mainly include 3-D point cloud reconstruction technology and 3-D workpiece assembly defect detection technology.The research contents of this dissertation are mainly from these two parts.Firstly,aiming at the 3-D reconstruction technology,the mainstream 3-D reconstruction methods are investigated.Considering the reconstruction accuracy and speed,the coding structured light 3-D reconstruction method based on multi-frequency heterodyne is selected to obtain the 3-D point cloud of the workpiece.Hu moment invariants and NMI(Normalized Moment of Inertia)features are used to locate the unwrapped region accurately,and the unwrapped region is decomposed.In addition,in order to reduce the point cloud loss caused by step error in phase unwrapping,λ1’=1008,λ2’=144,λ3’=16 are selected as the three wavelengths of encoding structured light.The experimental results show that,compared with the traditional multi-frequency heterodyne 3-D reconstruction algorithm,the proposed Coded Structured Light 3-D reconstruction algorithm for specific areas improves the reconstruction speed by about three times,and can effectively avoid the point cloud missing caused by mismatching,thus realizing the high-precision and fast 3-D collection of point cloud data of industrial components.Secondly,for the 3-D defect detection algorithm,by comparing the existing 3-D point cloud denoising algorithms,the average curvature of point cloud is introduced into the 3-D point cloud denoising process.The point cloud to be processed is divided into flat areas and information-rich areas.Statistical filtering is used for smooth areas,and bilateral filtering is used for information-rich areas.In the process of feature extraction,the FPFH(Fast Point Feature Histograms)features of feature-rich regions are extracted as the basis of point cloud segmentation and registration,while the geometric features of flat regions are mainly collected.In order to make up for the shortcomings of the existing 3-D feature extraction methods,this dissertation uses the two-dimensional feature mapping method to supplement the 3-D features of the measured object.By determining the position of corner and edge which are easy to extract in the two-dimensional image,and using the result of binocular system calibration,the feature is transformed into the 3-D coordinate system,which enlarges the existing 3-D feature,and improves the accuracy and speed of defect detection of workpiece.Finally,in order to verify the effectiveness of the proposed algorithm,a hardware system for defect detection of industrial parts assembly is built,and the measurement accuracy of the system is evaluated.The experimental results show that the calibration error of the system is 0.016 mm,the average measuremcent accuracy of the system is 0.0598 mm,and the uncertainty of class A measurement of the system is 0.018 mm,which meets the needs of industrial measurement.In addition,the air switch assembly state detection and the box socket and hanging ear assembly effect detection are used as two specific cases in the 3-D defect detection of industrial parts assembly.The experimental results show that the proposed 3-D inspection method for assembly defects of industrial parts can identify defects and calculate the degree of defects,and has a certain robustness,and achieves the goal of 3-D inspection for assembly defects of industrial parts.