【作者】 陈少强；

【导师】 吴信才； 宋利好；

【作者基本信息】 中国地质大学（北京） ， 地图制图学与地理信息工程， 2003， 博士

【摘要】 三维地质建模理论的研究是3DGIS的一项基础性工作，三维地质模型是3DGIS的重要组成部分，是研制三维地学可视化软件和三维地学信息系统的基础和核心。 由于地质现象的复杂性，三维地学模拟的理论方法和技术仍处于研究阶段，还有许多问题有待解决。目前国外在采矿和地球物理领域的三维地学模拟已进入实用阶段，有了一些较成熟的软件，但国内尚无相应水平独立版权的三维地学模拟软件。 不同的建模方法决定了不同的数据结构及可进行的操作，本次论文将现有的地质模型分为两大类：体元充填模型和边界模型。体元充填模型用于描述非均质地质体，边界模型用于描述均质地质体。 实际的地质体都是复杂的、非均质的，在对地质对象进行模拟时，需要对其进行简化。目前，通过使用一定数目的均质模型来表示非均质地质体是一种实际、有效的手段。 本次论文使用的模型是用来表示均质地质体的边界模型，因为本次论文完成软件的侧重点是在模型的编辑方面，而表示非均质地质体的体元充填模型极不适合人机交互的情形。 “八五”期间，我们完成了PandaCAEX计算机辅助勘查系统的研制。这个系统使用多面体表示的边界模型，成功地实现了三维地质模型的计算机交互建模及人机联作重磁反演，达到了国际先进水平，并获得地矿部科技进步二等奖。这个系统对三维地质建模的主要贡献，是针对三维地质模型在交互修改时的困难，独创了一种所谓的“橡皮膜技术”。使用这种技术可以灵活、交互地构制任意形状的三维地质模型。 PandaCAEX的设计目的是解决金属矿的重磁正反演问题，而金属矿多是一个个孤立的地质体，所以模型算法是针对设计单个复杂的地质体的目标设计的，这在当时被认为是够用了。因此这种算法没有处理相邻模型的能力，相邻的模型间会产生不应有的重叠现象。 但地质体不可能都是分离的，不但金属矿有相邻的情形，更重要的是另一类地质体——层状地质体的存在。PandaCAEX系统完全没有处理层状地质体的能力，因为层状地质体其实就是一层层紧密相邻的地质体。 本次论文的研究目的就是解决PandaCAEX系统的相邻模型重叠问题，它独创了一种新的建模技术，既可以表达块状的金属矿模型又可以表达层状模型，重要的是它完全避免了重叠现象的发生。 这种算法的关键是在所定义的空间中，建立了一个辅助的三维网格，通过网格建立起模型之间的联系，以消除模型的重叠现象。其中的关键技术是多面体模型的光栅化和三维光栅数据的多面体化。多面体的光栅化是二维多边形光栅化的直接推广，本次论文的关键创新是解决了三维光栅数据的多面体化问题。 Marching Cubes技术是科学可视化中用于从三维规则数据场中抽取等值面的技术，本次论文对其进行了改造，将其用于三维光栅数据的多面体化，从而实现了从多面体到光栅，再从光栅到多面体的循环，为解决多面体模型的重叠问题提供了算法基础。 本次论文的成果是以本次论文提出的算法为基础，完成了一个人机交互三维地质建模系统。这个系统使用Visual C++从底层开发，具有良好的用户界面，用户可以从多个窗门中从不同角度同时对模型进行编辑。利用这个系统用户可以借助不同种类的数据，在所定义的三维空间中，交互地对各种复杂的地质对象进行模拟，包括模型的相邻、交叉、层状地质模型及断层。事实上在所定义的网格分辨率的精度下，理论上用户可以构造出任意复d一内地质体。 本论文对此软个1八旬小儿和限制条个1。进行了了详细的说明。并指山了今后的改进力一向。为例寺理论的完整性，本次论文还对二：维地质建模中攸川的一些关键技术，如：约人三角剖分技术；离散光滑插值 n引）技术；非均匀有理Bfy条（＊＊＊BS）技术等进行了较详细的论述。更多还原

【Abstract】 The study of the theory of three-dimensional geological modeling is one of the fundamental works in3DGIS. for 3D geological modeling is a very important component of 3DG1S. It is the foundation andkernel of 3D geosciences visualization software and 3D geosciences information system.Due to the complexity of geological conditions, the theory and techniques of 3D geological simulationare still in the phase of research, many problems are still unresolved. These days, in the field of miningand geophysics, 3D geological simulation has entered the practical stage in the west, having someexperienced software. But in our country there is no such software of our own.Different methods of modeling decide different data structure and the operations that can be carried out,this dissertation classifies the existed geological models into two general types: the volume-filling modeland the boundary model. The volume-filling model is used for the heterogeneous geological bodies andthe boundary model used for the homogeneous geological bodies.In fact, all of the geological bodies are complicated and heterogeneous. While simulating geologicalobjects, simplification must be done. At present, using some homogeneous geological bodies torepresent the heterogeneous geological bodies is a practical and efficient method.This dissertation use the boundary model representing homogeneous geological body, since the softwarefinished this time is focused on the editing of model and the volume-filling model representingheterogeneous geological bodies is completely not suitable for this situation.In the period of 8n 5-year-plan, we finished a software of computer aided exploration system namedPandaCAEX. This system used the boundary model represented by polyhedron, and successfullyachieved the interactive modeling of 3D geological model and the inversion of gravity and magneticanomalies. This system reached the advanced international standard in some aspects and acquired the 2ndlevel award of science and technology from the ministry of geology. The main contribution of thesystem to the 3D geological modeling is the original creation of the technique so called "rubbermembrane" aimed to overcome the difficulty of interactive modeling of 3D geological model. Using thistechnique, users can easily make arbitrary shaped 3D geological model interactively.The design target of PandaCAEX is to carry out the inversion of 3D gravity and magnetism of metalmine model, so the modeling algorithm is aimed to animate the isolated complex geological body, at thattime, that is thought to be OK. Therefore this modeling algorithm has no the ability to deal with adjoinedmodels, the adjoined models will has unwilling overlap.But not all of the geological models are isolated, the metal mine model has adjoining situation, and themore important situation is the existence of layer geological body. PandaCAEX system has no anyability to deal with such situation, for the layer geological body is some adjoined geological bodyindeed.The purpose of this dissertation is solving the overlapping problem in PandaCAEX when modelsadjoined, it creates a new modeling technique that can represent both the metal mine model and the layermodel, the most important is that it completely solve the overlapping problem.The key of the algorithm is the setup of an accessory 3D grid in the defined 3D space. Models can havesome relationship depending on the grid, resulting that the overlap problem be solved. The coretechniques are the rasterization of polyhedron and the polyhedronization of the 3D raster. Among these techniques, the rasterization of polyhedron is the direct extension of the rasterization of polygon in 2D space, the major creation of this dissertation is the solution of the polyhedronization of the 3D raster. Marching Cubes is the technique that extracts isosurface from 3D raster in scientific visualization. This dissertation modifies it, and then uses the modification to ma更多还原