Research On Construction Of Single Building Of Multi-View Image Matching Dense Point Cloud

The core idea of the 3D model construction based on Oblique Photography Technique is to build triangular grid model based on image matching point cloud which has high degree of automation,advantage of low cost,but the modeling principle of oblique modeling results in the following problems:The whole models of the region fails to achieve the model to be single,data of models is too large and part of the models data is distorted.Based on the principle of oblique photography modeling,this paper proposes a method for realizing construction of single building from view image matching point cloud,which solves the problem of data redundancy,distortion and specialization.The main work is as follows:(1)Firstly,the key technology of image matching dense point cloud is introduced,it is clear that camera calibration,aerial triangulation and image dense matching to improve the accuracy of matching point cloud.In order to most UAV can not be equipped with high-precision POS system,tilting-wide angle camera can improve aerial triangulation precision,therefore,the choice of tilting-wide angle camera,design a method that obtaining high precision intensity 3D point cloud by designing tilt photogrammetry,to ensure the density of dense three-dimensional point cloud.(2)The whole point cloud according to the image projection plane is divided into different points piece according to t the perspective of the relationship between image matching point cloud data,and the three-dimensional network problem is transformed into two-dimensional network structure,which can be used to construct a large number of TIN accurately and effectively.Based on the generalization of filtering and the accuracy and efficiency of the algorithm,a filtering algorithm based on normal vector region growth segmentation TIN is proposed,which can be adapted to various terrain and effective acquisition Ground points and the top of the building,the ground point interpolation generated DEM.(3)According to the boundary property of roof TIN,the algorithm of tracking TIN boundary is used to extract the rough contour of the building which is regularized based on the key point methods,combined with the extracted DEM to realize DEM supported rapid modeling for single white building.(4)In order to the problem that the traditional texture mapping steps are cumbersome and the degree of automation is low,the imaging conditions of the perspective geometry are completely discarded.The geometric white and texture images are taken from the multi-view image,and the image registration is not required.This paper designs a method of building texture rapid reconstruction based on multi-view image,which through the digital photogrammetry theory,based on the collinear equation,multiple images are automatically added to the known building white mold surface texture.The experimental results show that this research based on oblique photogrammetry-technology city building 3D reconstruction process and method,which has the characteristics of high degree of automation,vivid effect,low cost,to realize of single building and meet the LOD3 building model making conditions,to meet the need of practical engineering application which can meet the requirements of LOD3 class building model.