多源海量地理栅格数据库引擎技术研究

发布时间：2018-04-10 09:27

本文选题：栅格数据　切入点：多源　出处：《北京建筑大学》2013年硕士论文

【摘要】：地理栅格数据是GIS中最重要数据源之一，GIS的应用和服务越来越多地依赖于栅格数据类型。栅格数据具有数据种类繁多、数量庞大、数据格式复杂、生产速度快等特点，给数据管理和分发工作等带来挑战。目前很多部门使用基于文件管理地理栅格数据的方式，当数据量超过一定规模后暴露出一系列问题，严重影响到数据的精确检索、高效分发和利用，降低了数据的利用率和管理效率。针对以上问题，本论文首先从理论研究的角度，对多源海量栅格数据从其基本特征到空间数据模型再到空间数据结构进行了详细分析，从体系结构、栅格数据模型和空间索引三个方面对当前主流的空间数据引擎进行了研究和对比，，重点探讨了数据库引擎的体系结构、栅格数据存储模型、数据库系统设计、空间索引等方面的内容；在分析栅格数据库引擎的关键技术基础上，结合开源对象-关系数据库PostgreSQL技术设计实现了海量栅格数据库引擎GeoRDE。该引擎在多个科研项目中进行了应用实践，并取得了较好的效果。通过实践表明，本文设计的GeoRDE具有较高的数据存储访问效率、高效的查询检索、严格的数据安全机制、支持多用户并发访问等显著特点。本论文的主要工作包括以下几个方面： (1)基于栅格数据以及栅格数据库引擎的概念，探讨了栅格数据库引擎的功能、特点及研究主要内容，分析对比了几种典型的空间数据库引擎产品。 (2)深入研究了地理栅格数据模型和主流的空间数据模型，提出了多源海量栅格数据一体化存储管理的总体思路和理论框架，在此基础上设计了海量栅格数据库引擎的总体框架，以适应海量地理栅格数据的存储和管理；在探讨栅格数据储模型相关理论的基础上，设计了海量栅格数的空间数据存储结构，并建立了多源海量栅格数据库存储模型；详细分析了R-Tree和GiST空间索引，结合GiST数据库索引模板技术，设计了高效的栅格数据空间索引机制，以提高海量栅格数据的查询检索效率；为提高栅格数据网络传输的效率，设计了栅格数据压缩策略和异步传输机制。 (3)对比分析了海量地理栅格数据库引擎中的关键技术。分别从栅格数据组织模型及调度、空间索引机制、海量地理栅格数据的I/O优化三个方面进行了深入的研究。提出了栅格数据分层分块、异步传输、多级缓存的技术路线和实现思路。 (4)根据设计方案，在海量地理栅格数据库引擎的数据组织、关键技术、实现方法等的研究基础上，基于PostgreSQL数据库技术设计实现了栅格数据库引擎，并从数据调度、网络传输、安全机制等方面对GeoRDE核心技术给出了完整的实现。最后，结合课题研制，将GeoRDE应用到多个科研项目中。研究测试结果表明，自主设研发的栅格数据库引擎GeoRDE满足多源海量栅格数据管理的需求。
[Abstract]:Geographic raster data is one of the most important data sources in GIS and its application and service depend more and more on raster data types.Raster data has the characteristics of various kinds of data, huge quantity, complicated data format and fast production speed, which brings challenges to data management and distribution.At present, many departments use the method of managing geographic raster data based on documents. When the data amount exceeds a certain scale, a series of problems are exposed, which seriously affect the accurate retrieval, efficient distribution and utilization of data.The efficiency of data utilization and management is reduced.In view of the above problems, this paper firstly analyzes the basic characteristics of multi-source massive raster data from its basic characteristics to spatial data model and then to spatial data structure from the perspective of theoretical research.Three aspects of raster data model and spatial index are studied and compared. The architecture of database engine, the storage model of raster data, the design of database system are discussed in detail.Based on the analysis of the key technologies of the raster database engine, a massive grid database engine, GeoRDE, is designed and implemented with the open source object-relational database (PostgreSQL) technology.The engine has been applied in many scientific research projects and achieved good results.The practice shows that the GeoRDE designed in this paper has high efficiency of data storage and access, efficient query and retrieval, strict data security mechanism, and supports multi-user concurrent access.The main work of this thesis includes the following aspects:Based on the concept of raster data and raster database engine, this paper discusses the functions, characteristics and main contents of raster database engine, and analyzes and compares several typical products of spatial database engine.(2) this paper deeply studies the geographic raster data model and the mainstream spatial data model, and puts forward the general idea and theoretical framework of integrated storage and management of multi-source and magnanimous raster data. On this basis, the general framework of the magnanimous raster database engine is designed.In order to adapt to the storage and management of massive geographic raster data, the spatial data storage structure of magnanimous grid number is designed on the basis of discussing the relevant theory of raster data storage model, and the storage model of multi-source magnanimous raster database is established.The spatial index of R-Tree and GiST is analyzed in detail, and an efficient spatial indexing mechanism of raster data is designed in combination with the technology of GiST database index template, in order to improve the query and retrieval efficiency of massive raster data, and to improve the transmission efficiency of raster data network,The raster data compression strategy and asynchronous transmission mechanism are designed.The key technologies of massive geographic grid database engine are compared and analyzed.The organization model and scheduling of raster data, spatial indexing mechanism and I / O optimization of massive geographic raster data are studied in detail.The technical route and realization idea of hierarchical block, asynchronous transmission and multilevel cache of raster data are put forward.According to the design scheme, based on the research of data organization, key technology and implementation method of massive geographic grid database engine, the grid database engine is designed and implemented based on PostgreSQL database technology, and the grid database engine is designed and implemented from data scheduling, network transmission, etc.The security mechanism and other aspects of the core technology of GeoRDE is given a complete implementation.Finally, GeoRDE is applied to many scientific research projects.The experimental results show that the Raster Database engine (GeoRDE), developed by ourselves, can meet the requirements of multi-source and magnanimous raster data management.
【学位授予单位】：北京建筑大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P208

【参考文献】