Research On The Methods Of True Three-Dimensional Mesh Construction Of Urban Point Cloud

Detailed mesh models are urgently required during digital city construction,and urban point cloud can obtained from the oblique photogrammetry.Under such a background,urban mesh construction has to be researched.The paper dives systematically into such techniques as preprocessing of discrete points,construction and simplification of true three-dimensional mesh etc.The work completed and accomplishments achieved in this paper are given as follows.1.The research works of mesh construction and mesh simplification at home and abroad are systematically summarized,which provide groundwork for subsequent true three-dimensional mesh construction.2.The preprocessing methods are introduced for urban point cloud with the features of discrete,intensive and uniform.The spatial index methods of space grid,octrees and k-d tree are described to improve the processing speed of discrete point cloud.The experiments show that k-d tree is the most efficient method for large amounts of data,and the search efficiency of the three methods is similar for small amount of data(one million or less),in addition,grid method has certain advantage for searching within a radius.In order to remove data redundancy,a uniform simplify method is described and implemented.The simplified results can retain the contour feature in the original point cloud.3.As the models obtained by traditional methods lack of side information,a mesh construction method suitable for urban point cloud is presented.The method generated urban mesh model by the steps of mesh initialization,mesh growth and holes patching.Double constraints are applied to improve the search efficiency of point cloud.Topology relation of triangles is discussed to achieve the mesh growing.Intersection angle and triangular holes patching are used to optimize the mesh.The experimental results show that the mesh constructed by the proposed method can exactly describe the city entities including buildings and bridges.With the mesh of compressed point cloud,the application of this method is analyzed.4.For the requirement of LOD models in 3D visualization,half-edge collapse simplified methods are discussed based on quadratic distance,area weighted quadratic distance and volume square.The experiments show that the area-weighted quadratic distance error method,simplified the mesh well,can used to achieve multiple level of detail model for urban mesh.In view of the phenomenon of poor quality triangles in the simplified mesh,a mesh optimization based on Laplacian smoothing and edge exchange is proposed.The results show that the method can improve the triangular quality in the case of small geometric errors.