Research On Construction And Optimization Of 3D Model Of Oblique Photography

At present,oblique photogrammetry technology is a relatively effective method for constructing 3D models of large scenes.The model results obtained are highly accurate,contain rich geographic information,and can truly restore the real world,and this technology has been widely used Smart city construction,geological disaster monitoring,smart tourism industry and other fields.Although the traditional modeling method can obtain a more refined 3D model,it is not suitable for 3D modeling of large scenes.Oblique photogrammetry uses a drone equipped with a multi-lens camera to efficiently collect multi-view images and quickly build a real threedimensional model.In the process of building a 3D model by oblique photography,the 3D reconstruction technology of oblique multi-view images is currently a problem that has been widely concerned.This article first systematically sorts out the concepts and principles of related technologies including image matching,beam network aerial triangulation,and dense matching of multi-view images.Take the actual project as an example to construct and optimize the large-scale 3D model.First,carry out image matching experiments through different algorithms to compare and analyze the effect of image matching.In response to the inevitable model voids,texture deformation,suspended matter and other problems in the 3D model results constructed by oblique photogrammetry,several modification experiments were carried out,and a more effective and fast model optimization scheme was concluded,which can realize refined model construction.The main research work and conclusions of this paper include:(1)The construction of the theoretical system of 3D reconstruction of oblique photography.Through reading a large number of related documents,the relevant concepts and principles of3 D reconstruction of oblique photogrammetry have been clarified,mainly including image matching technology,beam method aerial triangulation,and dense matching of multi-view images.Then through the in-depth study of the more popular SIFT algorithm and SURF algorithm in image matching,by selecting two sets of continuous oblique images with different characteristics to conduct experiments,the test results of the two algorithms are compared and analyzed.From the experimental results,the SURF algorithm is more Suitable for continuous tilt image matching.Finally,the common algorithms for dense matching of multi-view images are summarized,mainly including motion recovery structure algorithm,multi-view clustering algorithm and so on.(2)The optimization practice of the 3D modeling process of oblique photography.Take a certain area in Jiangxi Province as the research area and use the ContextCapture platform for3 D modeling.Aiming at the air three tasks in the modeling process,after many experiments,a user connection point optimization scheme is proposed,which can improve the quality of the air three results,reduce the number of times of submitting the air three tasks,and save the overall modeling time.The complexity and geometric accuracy of the three-dimensional model constructed by oblique photography are studied and analyzed according to the standard system,and the model results fully meet the requirements of the standard.(3)Research on optimization of 3D model of large scene.The defect types and causes of the three-dimensional model are systematically studied,and corresponding modification methods are proposed for different defects.The problem model is modified and replaced to achieve the purpose of model optimization.It also proposes a method of extracting tiles from the problem model area based on the intersection analysis function of ArcGIS software,which solves the problem that there are too many tiles in the 3D model of a large scene to quickly select defective tiles,which greatly improves The work efficiency of model optimization is of reference significance for constructing refined models in actual work.