Construction Of The Main Building Model Based On Point Cloud Dat

Driven by digital cities and smart cities,people’s spatial cognition of urban 3D scenes is constantly improving,and the demand for urban 3D models is also increasing.Buildings are the basic elements of urban space,and their identification,extraction,and information update play an important role in the needs of people’s daily life,urban planning and design,and urban economic development.With the continuous development and progress of sensor technology,modern sensors already have the advantages of diversification,high sensitivity,high resolution and high precision.How to combine people’s actual production and living needs and meet the intuitiveness,accuracy,Integrity,obtaining 3D models of buildings and even 3D models of cities from high-precision,high-resolution 3D point clouds is one of the current hot spots.However,most of the existing 3D model reconstruction methods rely on human interaction,and the extremely high manpower and time costs are not conducive to the reconstruction and application of large-scale urban models.Aiming at the problems of high data source complexity,lack of structural information,lack of multi-moderate constraints,and large amount of data in the process of informational expression of buildings in the existing 3D urban model.In this paper,the individual reconstruction of the building main structure model is taken as the goal,and the feature extraction,structural expression and model simplification of the building are carried out from the perspective of feature-driven model.Based on the theory of graphics and images,the construction method of the building structural main structure model based on line features is proposed Finally,the feasibility of this method is verified by experiments.The main work and research content of this paper are as follows:(1)Domestic and foreign research status and 3D reconstruction related theory introduction.Starting from the expression and practical application of the 3D model of buildings,the main research background and significance of this paper are firstly described,and the existing research status of building feature extraction,3D reconstruction and mesh model expression based on point cloud is summarized.Introduced the basic features related to the building,the basic concepts of computational geometry and the expression of the 3D model,and determined the main research content and overall thinking framework of this paper.(2)Building main structure reconstruction based on Hierarchical Strategy.The main idea is to extract multi-level information of the building from hierarchical features,the multi-level features of buildings were extracted from the sliced point cloud,then use simple rectangular primitives for building feature expression,and construct the building wireframe model through robust connections of multi-level rectangular primitives.(3)Reconstruction of building structures based on multi-view projection.To address the problem of low-quality extraction of building point cloud features,a building feature line extraction method based on multi-view projection was proposed.On the basis of extracting the building feature surface,this method compensates for the missing situation of the building point cloud from different perspectives,and solves the problem of integrity and coherence between building line segments.Then,in order to obtain the main frame structure of the building,the selection of the main line segment based on graph cut optimization is proposed,which provides a new solution for the subsequent 3D reconstruction based on the main line feature of the building.