面向云计算环境的泛知识化三维表达模型研究

Currently, three-dimensional data models have serious shortages in representing the complicated and dynamic three-dimensional world and in adapting modern computing environments. On one hand, the complexity, diversity and unpredictability of three-dimensional objects, makes it unrealistic to seek a universally applicable data model. On the other hand, the dynamic, high-performance on-demand cloud computing puts more emphasis on the integration of “storage and computation”. But traditional data models separate both storage and computation, thus unsuitable to the cloud computing. Therefore, on the basis of characteristics of human recognition and by taking knowledge granularity as basic units, heterogeneous structures (i.e. multiple representations including such as discrete, function and mapping structures) are to be normalized in this research, and a ubiquitous knowledgeable three-dimensional model that is suitable for the cloud computing is to be established; a knowledge database of shape semantics will be formed in an open and crowd sourcing mode, which has the ability of integration. This study is cloud-computing-oriented, and will break through the bottleneck of 3D GIS key techniques, boost the innovation and development of GIS in the description and analysis of three-dimensional world, reduce the cost of popularization and application, and improve the application efficiency. Thus application demands of multiple purposes (users) of 3D GIS can be flexibly and conveniently satisfied in cloud computing environments.

目前，三维GIS模型在表达复杂多变的三维世界和适应现代计算环境上，均存在不足：一方面，由于三维世界的复杂性、多样性以及不可预见性，寻求一个放之四海而皆准的模型是不现实的；另一方面，云计算动态、高性能的按需服务更加强调“存算”一体，传统分离的数据模型难以满足其应用需求。为此，本研究基于人类认知规律，以知识粒为单元，实现异质结构（即多种表达方式，如离散、函数、映射等）归一化表达，进而构建适于云环境的泛知识化三维表达模型，并构建开放、众包的形态语义知识库，使其具备兼容、进化和拓展能力；同时，基于“存算”一体要求，探索空间数据云存储基本原则与策略，以知识粒为单元建立空间数据自适应云存储技术。本研究面向云计算，将突破制约3D GIS关键技术的基础理论，带动GIS三维世界知识化表达发展与创新，更加方便、灵活地适应云计算应用需求。

目前，三维GIS模型在表达复杂多变的三维世界和适应现代计算环境上，均存在不足：一方面，由于三维世界的复杂性、多样性以及不可预见性，寻求一个放之四海而皆准的模型是不现实的；另一方面，云计算动态、高性能的按需服务更加强调“存算”一体，传统分离的数据模型难以满足其应用需求。为此，本研究在已有的（项目团队提出的）泛知识化三维表达理论基础上，基于空间对象运动行为，构建了时空统一表达的对象化数据模型，该模型具有灵活多样的结构（离散、函数、映射等），并通过使用对象与子对象的嵌套结构，充分纳入了空间对象的几何及语义层次特，因而可有效表达任意时空对象，提高了地理信息系统的建模能力。该模型实现了空间对象内部的几何可拆分性，提高了海量对象（对象内及对象间）的分布式空间索引效率；提出了可有效判定三维点包含关系的“点圆理论”，为三维求交计算奠定了重要理论基础，该“点圆理论”在二维求交计算中仍然有效，据此我们进一步突破了空间叠加计算这一GIS空间分析的复杂基础算法。这些技术适合于现代云计算及高性能并行计算环境，这为地理信息系统在智慧城市复杂与多样化应用奠定了重要理论基础。