Image Matching And 3D Modeling Of Close-range Photography-Assisted Tilt Photography

The UAV oblique tilt photography technology efficiently constructs a largescale,high-precision 3D real-life model,which can clearly and truly perceive the spatial positional relationship and texture structure information of reconstructed features in complex scenes.This technology is widely used in digital city construction.The restoration of ancient buildings and the monitoring of landslides.Image matching is a key step in reconstructing 3D models,and thus has become one of the important topics in the establishment of 3D models.In this paper,the development of UAV tilt photography technology,multi-view image matching and the construction of 3D model are analyzed.The three-dimensional modeling of tilt photography often shows the limitations of near-surface data loss,detail blur,and model deformation.It mainly captures a wide range of multi-view image data around the close-range photogrammetry-assisted drone tilting photogrammetry to achieve high-quality 3D model refinement.The main work of this paper includes the following three points:(1)Systematically expound the theoretical principle and system composition of photogrammetry,and introduce the key technologies in the process of tilting photography to produce real 3D models,including image matching of the same name,regional network adjustment,surface reconstruction and texture mapping,etc.And the subsequent production of urban real-world models provides a theoretical basis.(2)Research on image matching technology.The principle and implementation steps of SURF algorithm,SIFT algorithm and ASIFT algorithm are introduced.To measure the performance of three algorithms to extract feature points,image matching experiments with different perspectives are designed.The ASIFT algorithm is robust to affine changes.It is more suitable for feature extraction and matching of oblique images.It also proposes a calculation method for absolute tilt and relative tilt for multi-view oblique image matching.The experiment proves that the ASIFT algorithm feature matching result has a negative correlation with the tilt,which proves The rationality of the relative slope definition further applies the ASIFT algorithm to the matching of oblique images and close-range images.(3)Three-dimensional real-world model construction method and refined research.With the Smart3 D modeling platform,the real 3D scene reconstruction of a certain campus is carried out,and the complexity of the model results,the geometric precision and the accuracy of the plane model are evaluated and the error analysis is carried out.On the basis of the tilted photogrammetry reconstruction 3D model meets the requirements of the specification,The closerange photography ground image and the low-altitude ring-shooting image assisted the tilting photography to refine the three-dimensional model,and explore the fusion modeling of multi-source images,which makes up for the lack of nearsurface detail of the oblique photography and enriches the side texture information.