Research On High Precision 3D Terrain Model Construction Technology Based On Multi Source Data Collaboration

High precision Digital Elevation Model(DEM)is the most important basic geographic data to reflect the topographic features.In recent years,with the continuous acceleration of urbanization,the surface morphology has also shown diversified and discontinuous characteristics.At this time,the existing DEM cannot effectively express the real morphology of the surface,which seriously restricts the application of DEM in these areas.Although the emergence of various new data sources have improved the elevation accuracy of DEM,the transformation of the natural terrain into a combination of natural and artificial,abrupt terrain and continuous terrain,due to the transformation of the surface by human activities,is difficult for a single data source to meet the demand of high precision DEM construction,which will not only lead to distortion of modeling results but also cause the disorder of urban surface water catchment path.In view of the above problems,it is necessary to adopt a variety of new data sources with fast update efficiency and easy access,and propose a DEM construction method specifically suitable for the combination of natural and artificial,abrupt terrain and continuous terrain staggered distribution.The main research results and contents are as follows:(1)Modeling data processing and extraction: since precise ground points and accurate terrain boundaries are required to build DEM,this thesis processes and extracts the terrain data required for DEM construction,firstly using a mature filtering algorithm to extract the precise ground points required for modeling from the original Point Cloud;since the orthophoto covers a large number of natural terrain and common artificial terrain in life,this thesis classifies the orthophoto based on Deep Learning,and extracts the boundaries of various terrains efficiently and accurately to provide the data basis for subsequent construction of DEM.(2)Research on DEM modeling method of typical terrain: this thesis adopts the idea of "classification,modeling and fusion" to design different terrain modeling methods based on different terrain forms,elevation information and terrain functions.For roads and slopes,which are priority terrain for morphological expression,modeling methods are designed to take morphological features into account,which ensures the realistic morphology of roads and slopes;for the terrain with the priority of functional expression,such as square,a DEM construction method is designed to take into account the water catchment function;the road DEM constructed by the traditional interpolation method is modified by the correction algorithm to correctly simulate the water flow direction;finally,according to the boundaries of different types of terrain,different fusion methods are designed for DEM fusion to ensure the natural connection between terrains.(3)An example verifies the applicability of the method in this thesis.In this thesis,Chuzhou University is used as the study area,and each terrain modeling method is used for experiments.The experimental results show that the terrain model constructed by the modeling method in this thesis can meet the requirements of high-precision modeling,both in terms of elevation accuracy and morphological accuracy.For morphological expression priority terrain,this modeling method can better reflect the geometric semantic and morphological features of each terrain;for functional expression priority terrain,this modeling method can better reflect the functional features of the terrain.Therefore,the modeling method in this thesis can clearly express the frame structure of surface morphology and meet the standard of fine modeling.