【作者】 赵伶俐；

【导师】 朱建军； 赵仁亮；

【作者基本信息】 中南大学 ， 地图制图学与地理信息工程， 2010， 博士

【摘要】 中国目前已经进入城镇化快速发展阶段,迫切需要建立城镇发展动态监测系统,对城镇动态变化进行监测,规划资源,保护耕地,协调城乡发展。城镇化数据具有多源、异构、多尺度、数据量大、数据制式差异较大的特征,这些特征要求我们必须将这些数据进行集成、链接成一个整体,为城镇变化监测提供可比、可用和系统化的数据资源,对城镇化数据进行数据整合。论文以城镇化多源异构数据的索引建立为目标,提出对空间数据与非空间专题数据如何建立索引,如何进行联合查询,如何使专题数据涉及区域范围在图上进行快速显示等问题展开相关深入的研究,最后开发实验系统对研究的成果进行了验证与分析。论文的主要研究内容和成果如下：(1)分析当前的数据索引技术,对城镇化数据建立索引。城镇化数据可以分为空间数据和非空间的专题属性数据,由于不同的数据特点,对空间数据和专题数据分别建立相应的索引；专题非空间数据建立分类索引,空间矢量数据建立R树索引。针对R树索引数据重叠问题,提出基于行政区域的分层数据索引方法。(2)分析现有建立空间数据与非空间数据关联的方式,对实现城镇化两类数据的联合查询存在的问题。目前联合查询主要存在两种方式,一是将空间数据和非空间属性数据保存在同一个数据库中,并要求建库时就建立联系,数据变化时,联系可能要发生改变,更新、修改较为麻烦；二是利用空间数据中的对象属性信息(例如：地址编号、地理编码等),与非空间属性数据建立联系。前者要求必须在建库时就建立联系,而后者查询方式单一,查询信息很局限。城镇化非空间专题数据大多以文件的形式存在,空间数据缺乏与属性数据直接的联系,不宜采用上述两种方式建立联系。(3)分析城镇化整合系统中与位置关联的非空间专题数据(如地名,河流名)在地图上快速显示的问题。通常的显示方法是地名在地图的属性数据表中进行比较,找到该地名的记录,然后根据记录中的shape字段,在地图中进行显示。然而对于城镇化专题数据,有时缺乏相应的信息,需要采用新的方法来实现地名在地图快速显示。(4)提出一种以区域名和区域编号、区域MBR为主要内容的数据存储索引模型的数据结构—层次索引树,来实现两类数据的联合查询。这个层次索引树中结点的MBR内容与R树索引中的MBR相对应,建立起空间数据与非空间专题数据之间的联系；这种方法能扩大查询范围,并能够快速查询多源数据。(5)提出基于层次索引树的与位置关联的专题数据涉及区域在地图上快速显示的方法,方法的主要步骤包括三个步骤：首先根据专题数据对应的区域名,找到相应的编号；然后根据编号在层次索引树中寻找,找到最小的包含该区域名的结点；最后根据结点的MBR内容,在地图中进行显示。(6)开发实验原型系统,对文章所提出的方法进行实验与评价。实验在.Net平台,结合Arc Engine和Microsoft SQL Server 2005开发原型系统,验证城镇化数据索引的建立,数据的联合查询与查询结果的快速显示,并对文中所提出的算法进行时空复杂度分析。实验证明文章所提出的层次索引树的有效性,既可以实现联合查询,还可以实现与空间位置关联的非空专题属性数据的快速显示。本文提出的层次索引树存储索引模型是切实可行的,能实现数据的联合查询,查找更多的数据,并能解决城镇化过程中的图上的快速显示。建立城镇化动态监测系统,就必须对多源数据进行整合,整合系统的建立的关键之一就是要对入库后的数据进行快速查询,以及进行相应信息地图上的快速显示,层次索引树的提出为城镇数据整合理论提供了实用而新颖的思路。更多还原

【Abstract】 With the rapid development of urbanization in China, the amount of valuable data captured in different systems and different structures grows noticeably. At the same time, technical progress makes it possible to establish link among these different systems and different structures. As a result, the wish to monitor changes in urban areas and exchange, share these data arises and becomes more and more important. The combination of data done on different levels that we subsume under the term data integration requires resolving heterogeneities, which still poses research questions.The dissertation aims to establish indexes for the stored multi-source data, which aid to query useful information rapidly within urbanization integration system and realize the visualization of query results on the map rapidly, etc. Finally, an antitype system is developed to evaluate and test the method proposed by the dissertation. The main research programs and results derived are as follows:(1) The indexes for urbanization data is constructed based on the analysis of present index structures. The corresponding metadata table is constructed as non-spatial data stored in the binary format. The metadata information are utilized to classify the data and to establish the classified index respectively for the attribute data. An R tree index is established for the spatial data, and a kind of hierarchy data index based on administrative region is put forward to deal with overlay problem of R tree in the dissertation.(2) This dissertation analyses the two ways of relations establishment and combined query between spatial data and non-spatial data. The first one is to put these kinds in the same database with the relation established before data loading. Once data are changed, the relations will be changed subsequently which causes trouble in updating and modification the relations. The second one is that utilizing objects properties (i.e. address number. geographic number.) of spatial data to establish relation with non-spatial data. The drawback of former is that the relation should be established beforehand, and of later is that the query of information is limited. As the non-spatial data of urbanization are mainly available in the format of files and the direct information of relations are lacking, it is not suitable to adopt these two ways to establish the relations.(3) The dissertation researches into the rapid map visualization of non-spatial attribute data associated with location (e.g. place names, river names) in urbanization system. In general, the present approach of visualization is like that:Firstly, place names are compared and found according to attribute table, secondly find the corresponding records, finally visualize place names on the map. However, this method is not suitable for our situation, as urban place names lack corresponding information. It is urgent to adopt a new method to deal with the problem.(4) A kind of hierarchy structure-the hierarchy index tree, which is constructed mainly based on area name, area number and area MBR, is proposed to realize combined queries. The nodes of the index tree are corresponded with MBR, establishing the relation of the spatial data and non-spatial data. The approach could expand query range and search more multi-source data information rapidly.(5) An effective approach based on MBR is proposed for non-spatial attribute data associated with location to be visualized on the map rapidly. The approach mainly contains three steps, firstly the corresponding number is found by area name, secondly, the node containing smallest area is found by searching in the index tree, finally visualizing on the map by MBR content of the node.(6) A prototype system on data indexes establishment, data combined query and rapid visualization of query results is developed using.Net, Arc Engine and Microsoft SQL Server 2005 to examine the methods and algorithms presented in this dissertation. Experimental results show evidence of the reasonability and practicability of the index tree proposed by this paper, which not only could implement combined query, but also could implement rapid visualization for query results of non-spatial data with locations.So a conclusion is given without doubt that the hierarchy index tree structure is feasible and can greatly search more information in multi-source data. Urban data should be integrated to establish urbanization monitoring system, one of the keys is to query data and visualize the results on the map rapidly. The study of hiberarchy index tree reaches expectant goal and can provides a novel idea for urban data integration.更多还原