Segmentation And Boundary Extraction Of Building Facade From Mobile Laser Scanning Point Clouds

Various buildings in the city are important infrastructures of the city,and the rapid update of their information plays an important role in the reconstruction of the city's refined 3D model and intelligent management.The mobile laser scanning system can quickly obtain the point cloud data of urban buildings.which provides a new data source for the extraction of building information.Therefore,it is of great significance to explore a method of building point cloud facade segmentation and boundary extraction based on on-board laser point cloud data.The building facade segmentation and boundary extraction in the vehicle-mounted laser point cloud data mainly include:the building facade is blocked more and the point cloud is incomplete;the surrounding noise has a greater impact;the point cloud of each balcony facade of the high-rise building is not easy to distinguish.Therefore,on the basis of summarizing previous research methods,this paper proposes a method for building facade segmentation and boundary extraction in a vehicle-mounted laser point cloud.The main work content is as follows:(1)In-depth analysis of the geometric shape of the building facade point cloud in the two or three-dimensional space,the progressive segmentation method of the building facade point cloud based on the direction-constrained growth is studied.Based on the grid index,the algorithm designs a seed grid that distinguishes different types of facades;determines the direction constraint model according to the different seed grids,uses different models to constrain the growth of the point cloud of each facade of the building,and uses the RANSAC algorithm finely extract the facade point cloud after coarse segmentation,and complete the progressive segmentation of each main facade point cloud and balcony facade point cloud;(2)Aiming at the noise problem encountered in the segmentation process,this paper proposes a density clustering algorithm based on multi-scale grid to distinguish between noise and facade point cloud,which effectively solves the noise problem in the segmentation process;(3)For the divided point cloud of the building facade,this paper studies the method of boundary extraction using the linear projection features of the point cloud and the topological relationship between the surface feature structures,which can accurately extract the boundary points and lines of each building facade.Experiments on building facade segmentation and boundary extraction for different types of building point clouds,and compare the accuracy with the results of other methods.The experimental results show that the method in this paper can accurately segment different types of building facade point cloud data,the boundary extraction accuracy is high,and it has strong applicability and robustness.