Research On 3D Measurement Based On Multi-camera Vision Line Structured Light For High-speed Railway Train Body

The thesis aims at meeting the requirements of large-scale,high-precision,dense point cloud and high-efficiency measurement in the automatic grinding process of largecomponent high-speed train body.Based on the large-scale high-precision measurement technology of multi-camera vision line structured light,a large-scale high-precision measurement method for multi-camera vision line structured light based on auxiliary camera calibration is proposed to reduce the requirements for calibration tools and simplify the calibration process.At the same time,the global optimization algorithm is used to optimize the overall measurement system,reduce the point cloud data stitching error,further improve the accuracy of the measurement system,realize the global accurate measurement of the three-dimensional surface topography and provide high-speed rail body automated production technical support.The main innovations and research work of the paper include:(1)This paper proposes a new global calibration method for multi-camera vision line structured light based on auxiliary camera calibration,and establishes a large-scale highprecision measurement model for multi-camera vision line structured light based on auxiliary camera calibration.The global calibration is converted into the global calibration of the multi-camera vision system with overlapping fields of view,which solves the shortcomings of the high requirements of the calibration tools for the non-overlapping field of views of the multi-camera vision line structured light measurement system.The global calibration process can be completed with only one checkerboard calibration board,without the help of laser trackers,theodolites,etc.high-precision measurement equipment or complex targets,and the calibration process is simple and can achieve high accuracy.(2)In order to reduce the global calibration error inherent in the measurement system and solve the problem of misalignment of the point cloud data,a method of correcting the global calibration result based on point cloud registration is proposed.Only a common three-dimensional object is needed,and no high precision is required.For a three-dimensional target,a point cloud registration algorithm is used to solve the correction matrix for the rigid body transformation of the adjacent camera coordinate system,which optimizes the measurement system globally.Experimental data shows that the global calibration error of the measurement system is reduced from 0.0617 mm to 0.0446 mm after global optimization,and the global calibration error is reduced by 27.71%.(3)The trinocular vision line structured light measurement system was designed using five cameras.The experimental data shows that the accuracy of the line structured light unit of the measurement system does not exceed 0.036 mm and the global stitching accuracy can reach 0.0446 mm within the 790 mm measurement range.The measurement system scans and measures the high-speed train body side wall sample,and compares and analyzes the data collected with the Leica T-Scan AT960-XR handheld scanner.The overall error of the measurement system is 0.2982 mm.