Research And Application Of 3D Generation Technology Of High-resolution Remote Sensing Images

With the development and popularization of remote sensing technology,remote sensing images are more and more widely used in geological exploration,urban planning,disaster assistance decision-making and other fields.Terrain is one of the most important elements in remote sensing images.Accurate 3D terrain data is crucial in many applications.In remote sensing images,terrain can be represented by a digital elevation model(DEM).DEM is a mathematical model used to express the height of the ground.It is usually stored in raster data.DEM can be used for 3D terrain modeling,landform analysis,flood forecasting,etc.The data organization and expression forms of the digital elevation model include regular grid,triangulated irregular network and contour.In practical applications,the triangulated irregular network can better express the feature of terrain.The aim of this thesis is to improve the 3D terrain model generation algorithm and simplification of terrain algorithm in remote sensing images,so as to improve the efficiency of 3D terrain generation,simplification and rendering.The main research contents include:(1)A Delaunay triangulation network(D-TIN)generation algorithm that combines the method of very important points(VIP)and the Delaunator algorithm is proposed.D-TIN is used to replace the traditional regular grid to solve the problems of large data scale,data redundancy and inaccurate terrain representation.First,the algorithm uses a sliding window of the very important points method to calculate the difference of average elevation value between the center point and the eight surrounding points.Extract important points according to the difference of elevation values as D-TIN data source.The terrain extracted by important points can reduce data redundancy and retain the original features of the terrain.Then,the D-TIN is generated using the Delaunator algorithm.The Delaunator algorithm has low time complexity and high precision,and is suitable for large-scale 3D terrain generation.The experimental results show that the D-TIN generation algorithm proposed in this thesis has practical performance on terrain data of different scales,and can generate high-quality and high-performance terrain digital models.(2)A simplification of terrain algorithm based on the included angle between irregular triangular mesh surfaces is proposed.This algorithm uses the included angle of normal vectors between adjacent triangles to evaluate the complexity of the terrain,and sets the threshold to complete the simplification of terrain requirements under different fluctuations.Furthermore,in order to keep the simplified model undistorted,the meshing was used to keep the overall structure of the terrain unchanged.The simplification of terrain algorithm proposed in this thesis can effectively retain the detailed features of the original terrain.At the same time,the simplified 3D terrain model has high quality.(3)Finally,in this thesis,the Cesium is used to design and develop a 3D terrain display and application system,which can load and display the 3D terrain model generated in this thesis,and can be accessed on the Web.At the same time,the system is based on the Cesium framework and uses the 3D terrain model as the data base map,which further expands the practicability of the 3D terrain application.Through this system,users can perform operations such as browsing,roaming,and analysis on the 3D terrain model.The 3D terrain model can also be used as a basic data base in various fields,such as urban planning,traffic construction,emergency assistance,etc.The development of this system makes the 3D terrain model generated in this thesis more practical and valuable.