Single Building Extraction,Repair And Model Construction Based On Scene Point Cloud

In recent years,the rapid development of digital city and smart city construction promotes the development of 3D model,and the automatic construction of urban scene has become very urgent.Airborne LiDAR and oblique photogrammetry provide effective data acquisition means for the automatic construction of large-scale urban scene 3D model.However,airborne LiDAR and oblique photogrammetry are the surface data of urban scenes,which lack the semantic information of objects.It is necessary to further extract and model the individual objects of buildings.However,in the process of data acquisition,due to the occlusion of buildings(hidden parts)and other ground objects(such as vegetation),the building point cloud extracted from the point cloud can not cover the building surface evenly and completely,resulting in holes of different sizes,which brings great disadvantages to the complete modeling of buildings.How to obtain pure and complete point cloud data of single building is the basis of building 3D model construction.From the above considerations,this paper studies the extraction of single building point cloud data and the repair of missing areas.Based on the summary and analysis of existing methods,this paper focuses on the purification of single building point cloud data,the detection and repair of the boundary of missing areas.The main research contents are as follows:(1)Extraction and purification of single building based on scene point cloudIn this section,the 3D scene point cloud data is projected to the 2D plane to generate the point cloud feature image,and the single building data is extracted from the image edge detection method.This paper proposes a data purification method of building point cloud based on cloth simulation filtering,which can eliminate the non building data and achieve the effect of single building point cloud data purification.(2)Boundary detection of missing area in point cloud data of single buildingThe method of boundary hole detection based on grid is used to realize the detection and location of boundary hole;the method of edge feature detection is used to detect the edge information of the missing area;the method of vector deflection angle calculation is used to connect the edge feature points of the missing area into lines.(3)Repair and model construction of point cloud missing area of single buildingThe principal component analysis method is used to rotate the building to make the main direction of the building parallel to the coordinate axis;according to the building elevation information,the building is layered;the similarity judgment method is used to judge whether the building is similar between each layer;the repair method of missing area of building point cloud based on the structural characteristics of buildings is proposed to complete the missing area of building point cloud data Repair of the missing area.The rolling ball method is used to complete the 3D model construction of single building point cloud data.