Research On The Construction Method Of Refined 3D Model Based On Various Photography Methods

With the development of photogrammetry and computer technology,three-dimensional reconstruction has been widely used in digital cities,intelligent cities,cultural relics and archaeology.The upgrading of various applications makes the requirements for the integrity and refinement of 3D models increasingly improved.How to construct 3D models efficiently and accurately has become a research hotspot.The purpose of this paper is to construct a fine model which can preserve the integrity of large scenes and highlight local details.At present,the commonly used 3D modeling methods are divided into three categories:geometric modeling software technology,laser scanning technology,digital photogrammetry technology.In practical engineering,the 3D model constructed by using geometric software and laser scanning technology has high precision and good effect,but the method of constructing geometric software model has the problems of low working efficiency and long period.UAV low altitude tilt photogrammetry has high efficiency and low cost in obtaining large area data Many advantages,can quickly build a large range of real scene 3D.However,due to the limitation of UAV flight altitude and aerial photography acquisition angle,the structure and texture of near ground and occluded area are lost or insufficient.Aiming at the local detail problem of large scene scene 3D model constructed by tilt photogrammetry,the 3D model is constructed by combining close to photogrammetry,close-range photogrammetry and tilt photogrammetry.The main contents of this paper are as follows:(1)To construct a 3D model of large-scene oblique photography scene and to evaluate and analyze the accuracy of the model.Analysis and classification of buildings with distortion and drawing problems in the local area of the model.(2)The fine model of different types of ground objects is constructed by using various photography methods.The fine model is constructed by combining close photogrammetry with tilt photogrammetry for buildings with large size,regular structure and no tilt photogrammetry;for buildings with small size,complex structure and obstacles at the bottom,the fine model is constructed by combining close photogrammetry with close photogrammetry;for low buildings with complex structure,the fine model is constructed by close photogrammetry and then fused into large scenes.(3)To design an auxiliary scheme for the field image control point measurement and redundant image processing.Aiming at the problem of a large number of transition routes and redundant images taken during turning,the redundant images are automatically eliminated by judging the course angle and elevation change of the images.Based on the Python language,a low-altitude multi-dimensional remote sensing aggregation location aided design is realized in order to improve the field image control point measurement and change the established process of image control point measurement.Experiments show that the texture quality and accuracy of the model constructed by various photography methods can be greatly improved in fine modeling,and then the optimal scheme of fine modeling under different conditions can be obtained by comparative analysis.In the auxiliary design,the two schemes reduce the manual intervention and improve the working efficiency in the corresponding aspects.