High-precision 3D Reconstruction And Flaw Detection System For Large-size Targets

With the continuous development of large countries such as the high-speed rail industry and the large aircraft industry,large-scale machinery manufacturing technology has received more and more attention.The detection technology of large-sized and complex surfaces has become one of the important research directions in the industrial field.How to design a high-precision and high-efficiency detection system has always been an urgent problem to be solved.Today's technology is limited by the large size of large objects such as high-speed rail bodies or aircraft fuselages,and it is difficult to accurately and efficiently restore the three-dimensional appearance of the surface.Therefore,the subsequent target surface automatic grinding and automatic spraying are greatly affected.3D reconstruction technology is a core technology in the task of ensuring the smoothness of the work piece.At the same time,the defect area detection can also provide accurate position information for subsequent operations such as putty application.Therefore,this paper focuses on the design of a high-precision three-dimensional reconstruction and flaw detection system for large-scale targets,mainly for the upper and lower arc surfaces and side walls of high-speed rails,using line array sensors combined with three-dimensional reconstruction technology to achieve target surface topography reconstruction and flaw detection.The applicability and accuracy of the system were verified by simulation experiments.The high-precision 3D reconstruction and flaw detection system for large-size targets mainly includes:overall design of scanning scheme,target point cloud preprocessing,data registration,3D reconstruction,defect regional classification and detection.According to the project requirements,the sensor selection design contour array bracket is equipped with line array sensor to detect the target surface.According to the actual size of the work piece,this paper adopts a step-by-step scanning scheme and realizes 3D reconstruction through point cloud registration;The project requires the classification of possible defects and the development of target extraction criteria.After the target surface point cloud registration,the caliper-type segmentation measurement method is designed and combined with the least squares point cloud base surface fitting to extract the conformity extraction.The standard target defect area,at the same time,the key parts of the weld are separately extracted for subsequent targeted grinding and putty application.In order to improve the system detection and processing accuracy,a data acquisition error compensation mechanism is designed to improve the accuracy of the detection module.At the same time,the point cloud iterative closest point(ICP)algorithm is used to improve the point cloud registration accuracy.Finally,the simulation experiment is constructed.The 3D software is used to simulate the point cloud data of the car body surface.The ICP point cloud registration and 3D reconstruction are realized by C++programming,and the Mata surface fitting and defect extraction are performed by Matlab.Finally,the extraction result of the defect area is compared with the actual body data,and the processing precision of the system is analyzed.The experimental results demonstrate the effectiveness of the designed method.