Research On Point Cloud Precision Optimization Method Of Vehicle-mounted 3D Imaging System

In recent years,with the continuous acceleration of the digitalization process in China,the demand for geospatial three-dimensional information has increased sharply.As a brand-new data acquisition method,the vehicle-mounted three-dimensional imaging system has been used in urban construction,renewal,and transportation with its fast and efficient characteristics.Planning,smart agriculture and forestry are used more and more widely.High-precision point cloud data is the basis for the on-board 3D imaging system to stand out.However,due to system integration and the sensor itself,there are certain errors in the observed point cloud data.Therefore,the accuracy of the on-board 3D imaging system is improved.Research is crucial.This thesis takes the improvement of the point cloud accuracy of the vehicle-mounted 3D imaging system as the starting point,researches and discusses the composition,positioning principle,and error sources of the vehicle-mounted 3D imaging system,analyzes the influence of each error on the accuracy of the point cloud,and proposes the optimization of the vehicle's trajectory The main research content of the point cloud accuracy improvement method for consistency correction is summarized as follows:(1)The detailed study introduces the development status of the point cloud accuracy technology of the vehicle-mounted 3D imaging system,and summarizes the point cloud accuracy improvement research into three aspects: sensor-based,data-based and model-driven,analyzes the advantages and disadvantages of current methods,and clarifies the research goals and research content.(2)Explain in detail the composition principle and point cloud error of the vehiclemounted 3D imaging system,derive and study the positioning process and principle of the vehicle-mounted 3D imaging system,determine the key technology of system integration,analyze and study the error sources that affect the accuracy of the point cloud of the vehiclemounted 3D imaging system,and judge The degree of influence of each error on the accuracy of the point cloud is aimed at exploring ways to improve the accuracy of point cloud data.(3)Propose a smooth optimization method for motion trajectory,analyze and study the error of motion trajectory,screen the problematic motion trajectory in sections,and use the smooth optimization method to process the section trajectory within and between sections to enhance the data in the section.Consistency,enhance the data relevance between segments,use experiments to verify and analyze the rationality of the method.(4)A point cloud position deviation consistency correction method based on geometric feature matching is proposed.The revisited area is first detected.In order to avoid non-rigid deformation in the overlapping area,the point cloud data is refined and segmented according to the motion trajectory.,Use semantic segmentation to automatically identify and extract features to improve the automation level of data processing.Perform overall registration based on the obtained features of the same name,compulsorily correct the round-trip or multiple observation data in the same area,and improve the repeatability accuracy of point cloud data.