Research On 3D Building Model Reconstruction And Optimization Method Based On Line Festures Of Oblique Images

In recent years,with the gradual acceleration of China's modernization process and the advent of the era of big data,urban service systems such as "digital cities" and "real 3D cities" with geospatial information as their core are developing vigorously.Buildings are important infrastructures and components of urban areas and the construction of their three-dimensional model has became one of the key elements of the digital urban geospatial data framework.New challenges are posed to 3D modeling technology by high-level modeling requirements such as real-world 3D models,digital building models,and independent building models.How to construct three-dimensional models of urban areas,especially various complex buildings quickly,automatically and accurately is a hot issue in various fields.On the one hand,traditional vertical aerial photography or LIDAR radar can not obtain the building facade information,so they usually need to be supplemented with the digital close-range photogrammetry to build the building facade model,which is not only inefficient but also costly;at the same time,when building a 3D model of a building based on a point cloud,due to the huge amount of point cloud data,there are problems such as complex network structure,low modeling efficiency,poor model visual effect,and low modeling accuracy of texture missing areas.On the other hand,most of the existing modeling methods which are used to build 3D models are based on point cloud data and underestimate the role of line features.In the scenes of regularity and characteristics such as cities,since the three-dimensional line segment model is more effective in expressing the geometric structure of the man-made buildings,if line features are applied to the construction and optimization of the three-dimensional model of the building,the efficiency and accuracy of the model construction can be improved,the visual effects of 3D models can also be improved.Therefore,the reconstruction of building 3D models based on line features has important theoretical research value and social value.In view of many problems existing in the current building 3D model reconstruction field,this paper obtains the comprehensive data of the buildings by oblique photogrammetry,and based on the line features,key technologies in the building 3D model reconstruction technology route such as line feature matching,3D line feature reconstruction,3D model optimization,and rapid reconstruction of building models are studied.1.To solve the problem that the line features are difficult to match when the oblique image has large deformation and the occlusion is difficult,the matching robustness is not high when only the local features are considered,a multi-view matching method for oblique image line features under multiple constraints is proposed.A new idea and reference for the line feature matching of the oblique image and the multi-view matching of the line feature is provided by the solution.2.To solve the problem that the endpoints of the line segments are difficult to determine and the robustness is not high in three-dimensional reconstruction of line features,a three-dimensional reconstruction method of line features based on iteration method with variable weights is proposed,which solves the contradiction between line integrity and robustness in the 3D reconstruction of line features.3.To solve the problems of large computational complexity and poor modeling effects in texture missing areas for the buildings 3D modeling of based on point clouds,based on the line features of the building lines obtained by oblique photogrammetry,a reconstruction method of building3 D model based on the line features of oblique photogrammetry is proposed,which provides a new solution and inspiration for the rapid modeling of low-complexity buildings and the application of line features in the reconstruction of 3D models of buildings.4.To solve the problem of surface roughness and edge jaggedness in the construction of building3 D model of point cloud data,a three-dimensional model optimization method based on line feature is proposed,which improves the visual effect of the building 3D model and also improves the accuracy of the model.The method in this paper provides new research ideas for the optimization of the object triangulation and point cloud reconstruction model.