A Research On Key Technologies For Global Multi-resolution Virtual Terrain Environment

With the development of DE (Digital Earth) and GIG (Global Information Grid), how to implement a global virtual terrain environment has caught attention of scholars, and till now there are still many challenging problems in the air. Among these, is how to balance the overload between visual fidelity and real-time rendering, which is caused by global massive terrain data. To find a solution to this problem, we concentrate on some key technologies, such as designing a multi-resolution hierarchy for global massive terrain data, building an optimized terrain model and texture model, saving and indexing data in database, requesting terrain data of interest as far as possible from database during real-time rendering, representing the earth realistically as an ellipsoid, and at last trying to implement a sea simulation, which is quite difficult and having to be started with wave simulation at present.The followings are what we have done in this dissertation:1. Design a Pyramid Model for global multi-resolution virtual terrain environment, and then implement this model by way of MRLOD. Related to the model designing, is how to divide the whole world into different levels and blocks in terms of latitude and longitude. As for the structure of each terrain block, we introduce TIN model into the Pyramid Model, and explain why substitute TIN for RSG To solve the problem of texture minifying and magnifying and to speed up texture creation during real-time rendering, we adopt Mipmap texture method, which includes how to create Mipmap textures for each terrain block, how to compress these textures, and how to organize them in the Pyramid Model.2. In view of the importance and difficulty of building a TIN terrain model, we discuss it specially in a single chapter, explaining in detail data structures and key algorithms. The emphasis lies in how to optimize these structures and algorithms to better fit with triangulating and rendering. Another issue comes from cracks among blocks of different resolutions, which has been a long-time disadvantage of blocked terrain models. We put forward a new method to eliminate these cracks, which is quite easy and convenient for real-time rendering.3. Introduce the concept of DE (Data Engine) into VGTES, and discuss its and functions and relations with other modules. By comparing the two persist storage units, file and database, we choose the latter, which is better for saving and indexing global massive data. During the course of saving and indexing, we adopt the well-known method calledQuad-Tree, demonstrating how efficient it is in requesting blocked terrain data. Additionally, a COM-based middleware is built by virtual of ATL, just to access distributed heterogeneous database. An example is given to demonstrate how to access the Pyramid Model database, followed by main steps and VC++ codes.4. Introduce the concept of RE (Rendering Engine) into VGTES, and explain how close it is to the above Engine, Data Engine. Instead of traditional planar coordinate system, the GSRCS (Geodetic Spatial Rectangular Coordinate System) is used in VGTES to render the earth as a real ellipsoid, not a pseudo one, which cannot be measured accurately. But it is not easy to change from the planar coordinate system, for many new challenges will be confronted, such as how to measure distance and area on the ellipsoid, how to transform viewpoint around the ellipsoid, and how to cull view frustum to get a ellipsoid-based view field. All these are the key to VGTES, and hence become a major part of this dissertation. Another important part is view-dependent multi-resolution block algorithm, which will at last affect the efficiency of data requesting and real-time rendering. We find a solution to this problem, demonstrating its feasibility experimentally. In Addition, we bring forward a new method to project coordinates reversely from screen space to world space (i.e. geodetic coordinates). The reason why we abandon the project function provided by 3D API is because of the intrinsic disadvantage of z-buffer, which means the further from the observer, the lower depth precision.5. Try to implement a sea simulation since 70.9% of the earth is covered with seas. But it is quite difficult to simulate the whole sea because of its complexity, so we start with the most common phenomenon above sea surface, that is, sea wave. To simulate the wave, two points should be taken into consideration: the mathematic model and the lighting model. To the former, we use Perlin noise function and Octave function, and to the latter, we use GPU-based vertex shader and pixel shader. Groups of experiments are done to test whether or not these two models can meet the need of both visual fidelity and real-time rendering.6. Design and implement a prototype system named VGTES (Global Virtual Terrain Environment System). This system, developed by Visual C++, OpenGL, Direct3D, COM/ATL, HLSL, etc., is the result of above researches, and needs to be improved in the future.