A System For Measuring The Surface Topography Of Large-Size Workpieces And Research On Key Technologies

With the development of the industry,the requirements for the shape measurement of large-sized workpieces are constantly increasing,but it is difficult for the current general-purpose measurement equipment to meet various measurement requirements at the same time.Customized equipment is often required for different workpieces.This paper studies a topography measurement system for large-sized workpieces.The system adopts non-contact measurement technology and obtains the overall topography of the workpiece by means of point cloud splicing,which can efficiently and flexibly complete the topography measurement task.The main research work of this topic includes:Firstly,the measurement requirements are analyzed,and a topography measurement scheme for large workpieces is designed: The single-angle 3D point cloud measurement is carried out by using the binocular structured light technology,and then the point cloud rough stitching model combining the photogrammetric camera and the coded marker points is constructed.The target composed of marked points is fixed on the binocular measurement equipment,the target is photographed by a photogrammetric camera,and the point cloud data at each angle is spliced into the world coordinate system by the camera estimation method.In the aspect of single-angle point cloud measurement,a fast stereo matching algorithm is designed.The optimal matching point pair of the absolute phase map is quickly queried by the binary search method,and the edge matching error is reduced by using the sub-frequency constraint.The stereo matching speed is improved,and compared with the traditional BM algorithm,the matching speed is increased by12.34%.For the point cloud noise problem,a K-dimensional tree is used to construct the point cloud structure.On this basis,a density-based clustering algorithm is introduced.The number of binary connected domains of the absolute phase map is used to calculate the clustering threshold,and finally the isolated points and redundant point cloud blocks in the point cloud are effectively eliminated.In the aspect of multi-angle point cloud splicing,in order to improve the accuracy of point cloud splicing,the ICP algorithm is introduced to splicing point clouds of adjacent angles.Build topology between sets of multi-angle points.A global iterative point cloud splicing method is designed to complete the high-precision point cloud splicing of the overall shape of the workpiece.In order to quickly and efficiently extract marker outlines in complex scenes,a method for marker extraction based on multiple constraints is designed,which combines area perimeter constraints,coding ring constraints and corner constraints.Aiming at the problem of difficult calibration due to the large field of view of the photogrammetry camera,the self-calibration method is used to obtain the high-precision camera internal parameters through the unique constraint of the encoded markers,and then the cross-ratio invariance is used to calculate the distortion parameters.Aiming at the problem of difficult calibration due to the large field of view of the photogrammetry camera,the self-calibration method is used to obtain the high-precision camera internal parameters through the unique constraint of the encoded markers.Finally,the distortion parameters are calculated using the invariant property of the cross ratio.Finally,the system and test platform are built according to the design scheme,and the measurement accuracy of the system is verified by measuring blocks,pipes and large ring gauges.Experiments show that the point cloud measurement accuracy of this system can reach 0.073 mm,and the point cloud splicing accuracy can reach 0.153 mm,which can effectively complete the topography measurement task of large-sized workpieces.