Research On Construction Of Real-time Map Based And Point Cloud Classification On 3D Laser

In the context of the rapid development of electronic computers,3D laser scanners have made great breakthroughs in cost,efficiency and precision.At the same time,with the diversified development of scanning carriers,they have been used for such as ancient building protection and urban construction.Among the projects such as mold,topographic map drawing and BIM engineering,the high scanning precision and small environmental factors make the 3D laser scanning have long-term practical value in the construction of digital city.This paper focuses on the spatial characteristics of point clouds,and classifies urban buildings.In addition,this paper discusses a method of constructing real-life maps.The main research contents and conclusions of the article are as follows:(1)Point cloud data acquisition and pre-processing work to do a more in-depth study,including point cloud registration,point cloud noise reduction,point cloud streamlining,and a complete set of preprocessing operating.While dealing with the point cloud,this paper preserves the complete model features of the point cloud of the building and provides a basis for subsequent classification experiments.(2)Based on the point cloud cluster segmentation of Euclidean clustering,the model features of unit clusters are analyzed.Two spatial analysis methods based on normal difference DoN and principal component analysis PCA are combined with PCL source code and CloudCompare software.Cloud classification experiment.The results show that the progressive morphology can be used to cluster the point cloud better.In the DoN experiment,the normal vector field of the point cloud at the size scale is calculated.The two-norm value of the control vector can be used to segment the planar point cloud.In the PCA experiment,the sample point cloud is weighted by eigenvalues in multiple scale spaces,the feature distance is maximized to train the classifier,and finally the global point cloud is applied.The PCA classification method based on Euclidean clustering can ensure the integrity of building information and basically complete the extraction of urban buildings.(3)This paper projects the local 3D points to the 2D plane by normal projection,and combines the Delaunay principle based triangulation optimization algorithm to connect the 2D points and then map the 2D points to the 3D points.Surface reconstruction of 3D point clouds.At the same time,the spatial topology based on CSG-BREP is introduced to realize the spatial description from point to complex.(4)Building a three-dimensional model of buildings and ground point clouds based on Undet for SketchUp.In addition,this paper studies a city modeling software CityEnginge,including modeling process,CGA function method and experiment,road and plant construction methods,and finally uses the model replacement work to import high-precision building model model into the scene model,complete The construction of a real map.