Font Size: a A A

Automatic Reconstruction Of LoD3 City Building Model Based On Multi-source LiDAR Data

Posted on:2019-06-13Degree:MasterType:Thesis
Country:ChinaCandidate:Y LiFull Text:PDF
GTID:2370330545485813Subject:Photogrammetry and Remote Sensing
Abstract/Summary:PDF Full Text Request
With the rapid development of areas such as wisdom city,intelligent navigation and automatic driving,how to quickly get three dimensional space information of city buildings,build a high-precision three-dimensional fine model become a key problem to be solved.Based on the 2.5D features of airborne and vehicle-mounted LiDAR data,this paper establishes a set of technical solutions based on the data integration to generate 3d fine model using the 2.5D dual contour method.The specific research contents are as follows:(1)Study the method of building roof surface reconstruction based on airborne LiDAR data.Research is mainly focused on the roof surface reconstruction method of 2.5 D dual contour method,and the method is improved,considering the rules of the profile of the geometric structure of the building roof surface,this paper smoothes contour buildings using the algorithm of Douglas-Peucker to build triangular mesh model of the roof surface with fine details and rules on the edge,better reflect the real structure of the building.(2)Based on vehicle LiDAR data,the method of urban building facade extraction is studied.On the basis of the building roof surface has good reconstruction and regulation,using the adjacency of the urban building roof surface and the facade and the planar characteristics of urban building facade,this article using the method based on space domain to perform the fast clustering processing of vehicle LiDAR data to extract facade point cloud and RANSAC algorithm is adopted to define the facade plane.(3)This paper studies the combination of the roof surface model and the facade model of the urban building.After summarizes researches of domestic and foreign scholars in the field of point cloud registration,in view of the approximate orthogonal characteristics of the vehicle LiDAR data and the airborne LiDAR data,This thesis put forward a surface-to-surface matching strategy to realize the fusion of sub models.After the extraction of the the initial facade plane,we adjust the elevation parameters of facade plane according to the roof surface model,and adjust the level parameters of the roof surface model according to the facade plane,finally realizes the fine registration.Considering the city building facade and roof surface with 2.5 D characteristics,using 2.5 D double contour method generates city buildings facade model,the model with no apparent juncture,building roof face model can reflect the real structure of the building.Studies have shown that using the Douglas-Peucker algorithm to regulate building roof surface contour line can reflect the rules of the real structure of the buildings,but for buildings with complex contour may lose a lot of details;The method of spatial neighborhood can quickly realize the extraction of building facade.Based on the registration of face and face,it can realize the seamless fusion of the roof surface model and the facade model,which can reflect the real structure of urban buildings better.
Keywords/Search Tags:Multi-source LiDAR data, Facade extract ion, Registration, 3D modeling
PDF Full Text Request
Related items