Font Size: a A A

Monomeric Building Extraction Research On Dense Matching Point Clouds From Oblique Images

Posted on:2018-09-18Degree:MasterType:Thesis
Country:ChinaCandidate:Y ChenFull Text:PDF
GTID:2310330563951247Subject:Photogrammetry and Remote Sensing
Abstract/Summary:PDF Full Text Request
Multi-view oblique aerial photogrammetry can obtain ground object information from multiple views and realize three-dimensional(3D)reconstruction of targets with real texture and fine structure.Building model monomer extraction is an important research content in the field of tilting photography.The quality of the extracted results is of great significance to 3D modeling,model storage management,building attribute editing,3D visualization of urban buildings and so on.The paper conducts studies on the problems about monomeric building extraction on dense matching point clouds from oblique images as well as substantial modeling for monomer building point clouds.The main work and innovations are listed as follows:1.The technologies of oblique aerial photography and image data post-processing are discussed.The research statuses of 3D modeling and monomeric extraction of buildings are summarized systematically.2.A monomeric extraction method for building point clouds based on dimension reduction strategy is proposed.Firstly,point clouds feature image is generated according to the number and the plane and height distribution of clouds points in the plane grid with reduced dimension.And then the method on morphological threshold segmentation and Canny operator is introduced to extract the two-dimensional edge of the building.Finally,Pavlidis algorithm under the constraints of circumference and shape is applied to realize the contour tracking of the building,and monomer extraction for building point clouds is completed.3.Aiming at the characteristics of dense distribution and complex composition of urban buildings,another monomeric extraction method for building point clouds based on density clustering analysis is put forward.Firstly,ground objects and vegetation points are eliminated by elevation statistical method,and then DBSCAN clustering analysis algorithm is used to group the independent building models.Finally,the boundary of the building is determined by extracting the minimum circumscribed rectangle of the point clouds in the group,and monomer extraction of building point clouds without ground points is completed.4.A scheme combined with parametric model and the nonparametric model is used to construct the actual model of the monomeric building after extraction.The basic model library is established initially.Then,for building point clouds with simple structure,the building model is reconstructed by the parametric modeling method.For building point clouds with complex structure,the polyhedral structure model of buildings is established on the feature data acquisition and optimization method.Finally the texture mapping model is generated.The experimental results show that the scheme can realize the actual model construction of the monomeric building and is better applied to different types of buildings.5."Tilt photography building point clouds monomer processing software system" is independently developed,and the software has functions on point clouds read and display,monomer processing for building point clouds,monomeric building modeling and texture mapping of buildings and so on.The initial experiments prove the applicability of the software.
Keywords/Search Tags:Oblique Images, Dense Matching Point Clouds, Building Singularity, Dimensionality Reduction, Building Reconstruction, Texture Mapping
PDF Full Text Request
Related items