Three Dimensional Map System Based On Multi-Source Data

Three dimensional electric map(3D) is a three-simulation and abstract description of the real world according to a certain percentage. 3D map substitutes two dimensional abstract symbols for lifelike terrain and ground objects. It changed cartography theory from map transfer theory, map language theory and cognitive theory to geography cognitive theory, analysis of geographic data theory in 3D visual geography environment.The 3D electronic map research focuses on map data acquisition, map data structure, 3D map visualization and spatial analysis of 3D map. Map data acquisition is the basis of map research. It mainly consists of manual labor acquisition, airborne or aerial shot acquisition, etc. They are transferred to 3D modeling software for 3D modeling and then integrated into a 3D map system. Map formed by this way acquires not only a lot of man power, but also a long time. Moreover, it is hard to update data in maps. Data with different acquisition give map trouble on data fusion, data storage, data expression and topological relation description between 3D objects.Considering of the perspective of geographical factors, we proposed a method to generate a 3D map. The 3D map consists of 3D terrain and 3D buildings which is generated from both of remote sensing images and 2D plane map images. Multi-source data include Shuttle Radar Topography Mission (SRTM) Terrain data, multi-spectral remote sensing images and plane map images. We aim at implement a 3D map system of 3D roaming and editable functions by using graphics engine technology. Its realization depends on processing towards these three types data above.Research in the thesis includes terrain modeling based on SRTM terrain data, segmenting remote sensing and plane map images effectively, polygons recognizing from building contour, describing and saving map information to implement an editable map of terrain and buildings, providing a convenient interactive operation. What we have done and the achievements are as follows.(1)Designed an automatic generation map system based on multi-source data. Our system consists of five modules, namely, terrain modeling, building contour extraction and modeling, map rendering, map roaming and map editing. It contains three types of data which are SRTM data, multi-spectral remote sensing images and plane map images. Ogre and Qt are applied to render the map and design the user interface.(2)Improved the traditional watershed transform and gained a Multi-stage Watershed Transform based on Global Threshold(MWTGT) method. We segmented the remote sensing and plane map images by MWTGT and got a satisfy results.(3)Gained a mesh file to be rendering. Firstly, merging the segmentation results and then get its contour by canny algorithm. Secondly, collecting those collinear points and get the buildings'polygon contour by polygon picking algorithm based on least square method. Thirdly, subdivision those polygons.(4)Defined a XML file to describe and save map information. XML can be exchanged between different systems of different data formats. In the terrain modeling parts, SRTM data is changed into gray height image of PNG format. The XML and PNG image are delivered to our system as a base input of our map system.Experimental results show the performance of the algorithm and system. It is verified that MWTGT algorithm can reduce at least 88 percent of over-segmentation with maintaining performance compared to traditional watershed algorithm. And the 3D mapping system based on multi-source data is robust, high-performance and flexible.