| The Digital Elevation Model("DEM"),as a basic geographic information product,is a basic National Spatial Data Infrastructure(NSDI),which plays a huge role in many applications.At present,countries around the world are vigorously strengthening the construction of DEM databases,focusing on providing terrain data with good status,high accuracy and complete content.Since China entered the "Twelfth Five-Year Plan",the dynamic linkage updating of the national basic geographic information database has been put forward,aiming at continuously and dynamically updating the existing national basic geographic information database of 1:50,000,1:25 and 1:1000000.However,on the basis ofterrain details,under the large-scale product framework system,quickly updating the products in small scales and achieving multi-scale and multiresolution "high-fidelity" synthesis of multi-scale and multi-resolution DEM data on national and provincial platforms are the key technical problem to realize the dynamic update and linkage update of DEM data.For a long time,through the study of multi-scale comprehensive research on DEM data,it has been found that the precision of terrain expression and description and the quality evaluation standard of comprehensive results are the key and core issues of research.Due to the lack of sufficient accuracy and objective indicators,the relationship between the DEM comprehensive results and terrain scales are blurred.At the beginnig of this paper,a well-defined description system for typical terrain and landforms given out,which aimed to establishe a set of logical descriptions and digital expression systems for typical topography.Then the multi-scale definition and extension of traditional DEM estbalished from the spatial and attribute dimensions.After that the DEM multi-scale expression model formed by vector,grid and semantic combination.Which deducts the topographic feature and scale relationships to form the terrain factor expression of the terrain feature at different scales;next establishes the DEM digital synthesis algorithm under feature constraints through the effective extraction of terrain feature information,and solves the problem of DEM cross-scale highfidelity digital synthesis in the algorithm level from the perspective of rule constraints,.In summary,this paper proposes a multi-scale DEM digital synthesis technology under feature constraints.The main contents include:(1)The structural features such as topographic relief and texture pattern of typical topography at a certain scale are studied by analyzing the existing DEM libraries of different scales in China.A multi-scale DEM model for typical terrain and landform is construct by introducing spatial and attribute features to make vector geometry expression,grid texture expression and attribute semantic expression of terrain feature factors.Considering the application requirements,the directional profile characteristics of typical terrain and landform,as well as the point features,linear features and point and line hybrid features of line factor expression at multiple scales are further analyzed.The relationship between texture style and geometry representation of typical terrain and landform in multi-scale is established;consequently,the multi-scale terrain factors of typical terrain and landform are constructed.(2)Based on the quantitative statistics,information entropy calculation and spatial scale difference analysis of terrain texturefeatures,a multi-scale terrain texture factor extraction method based on deep learning is proposed.By marking the terrain and landform feature of typical high mountains,plains,hills and mountains,the learning model of multi-scale terrain and landform texture features is realized.On this basis,the SRCNN model has been applied to construct super-resolution reconstruction to realize the learning of the typical terrain and landform factors in different scales,which lays the foundation for constructing the rational features of the terrain for geometric synthesis.(3)A multi-scale DEM comprehensive constraint rule based on terrain texture feature factors is proposed.The information entropy constraint rules,geometric importance constraint rules,multi-directional section constraints and feature fusion filter constraint rules of topographic texture feature factors are constructed.The algorithm thresholds corresponding to the constraint conditions are refined.(4)A geometric multi-scale DEM synthesis algorithm based on topographic texture feature factor texture constraints is proposed.The DEM synthesis method based on triangular encryption points is realized under the point feature constraint corresponding to the texture features of different scales.The Douglas-Puke synthesis method under multi-directional section constraint and the DEM multi-scale synthesis under the hybrid constraint of point and line features are analyzed.The method realizes the geometric feature mapping of texture features and finally constrains the geometric integration of multi-scale DEM.Finally,based on the completion of the relevant model and algorithm design,this paper aims at the specific application environmentand the DEM data based on existing series of scales,including multi-resolution DEM under different scales,and different precision under the same scale.The DEM generated by the data source realizes crossscale high-fidelity digital synthesis.Based on the construction and update of China's national basic geographic information database,the data collection and update of the existing multi-scale database has been used to obtain the basic geographic information of the scale application applied to meet the needs of social and economic production and life.This paper realizes the cross-scale one-map data service,dynamically updates the existing database to build application norms,and provides strong data support for the national economic construction and production activities of various departments at all levels. |
PDF Full Text Request