Building modeling method under constraint of spatial topological relation based on LiDAR (Light Detection And Ranging)

Abstract

The invention discloses a building modeling method under constraint of a spatial topological relation based on a LiDAR(Light Detection And Ranging). The method comprises the following steps of carrying out point cloud segmentation processing and building contour extraction on LiDAR point cloud of a building in a target region to obtain an accurate two-dimensional outer contour of the building anda mathematical expression of geometric primitives of a building roof; constructing a spatial topological relation graph of the geometric primitives of the building roof; defining a three-dimensional Boolean operation rule between the geometric primitives of the building roof; carrying out three-dimensional Boolean operation on geometric primitive polyhedron of each building roof to obtain half-space of a primitive top surface, obtaining the half-space which completely contain wall surfaces and the ground of a building solid model according to the accurate two-dimensional outer contour and height information of the building, and carrying out an intersection operation of the three-dimensional Boolean operation on the half-space of the primitive top surface and the half-space of the wall surfaces and the ground to obtain the building solid model. The building modeling method under the constraint of the spatial topological relation based on the LiDAR has the advantage that the spatial breakage degree of a building model is reduced.

Description

technical field [0001] The invention relates to the technical field of airborne laser radar, more specifically, it is a building modeling method under the constraints of spatial topological relationship based on LiDAR point cloud. Background technique [0002] LiDAR (Light Detection And Ranging) is laser detection and measurement, which uses GPS (Global Position System) and IMU (Inertial Measurement Unit, inertial measurement unit) airborne laser scanning; the measured data is the Digital Surface Model (Digital Surface Model) , DSM) discrete point representation, the data contains spatial three-dimensional information and laser intensity information. Apply classification (Classification) technology to remove buildings, artificial objects, covering plants and other measuring points in these original digital surface models, and then obtain a digital elevation model (Digital ElevationModel, DEM), and at the same time obtain the height of the ground cover. [0003] The existing...

AI Technical Summary

Problems solved by technology

[0004] At present, there are two main methods for building model configuration: one is to use the step edge and roof line features extracted from the LiDAR building point cloud as the basis for space division, and reconstruct the building boundary based on the BSP algorithm. Reconstruct the building model; the second is to reconstruct the building model based on the "Split-Merge-Shape" algorithm, where the two steps of split-merge are mainly to reconstruct the topological relationship between the building structure lines, Complete the closed polygon; the two methods have certain similarities in principle, the core lies in the space division process, the correctness of the space division results directly determines the effect of the final model reconstruction; however, because the spatial topology between the patches is not restored in advance There is a certain blindness in space division, which is easy to cause broken small spaces, thus affecting the accuracy of the final building reconstruction results

[0005] The application number is 201110218035.9, and the patent name is "A Method for Quickly Extracting 3D Outline Information of Buildings from Airborne LiDAR Data", which proposes a method for quickly extracting 3D outline information of buildings in airborne LiDAR data. The patent titled "Method for Extracting Building Outlines from Lidar Point Cloud Data Based on Curvature" proposed a method for extracting building outlines from Lidar point cloud data based on curvature, but did not further give how to extract building outlines based on the extracted A Method of Constructing Building Model Based on Contour Information

Images