面向高效检索的多源地理空间数据关联模型

Multi-Source Geospatial Data Correlation Model for Efficient Retrieval

地理空间信息往往包含矢量数据、栅格数据和文本描述信息,这些信息之间通常相互联系.如何快速、全面检索和定位这些相关联的信息,是地理空间信息应用中的新需求.为提高地理空间数据检索和分析的性能,该文提出一种面向高效检索的多源地理空间数据关联模型MSGCM.该模型通过提取多源地理空间数据空间信息、语义描述信息、内容描述信息及其关联关系,构建特征要素图,并基于关联模式将多源地理空间对象融合到统一空间中.通过计算不同对象之间的关联强度,构建类似图的关联模型.为提高模型构建效率,提出了一种基于特征索引的分块构建方法.与已有方式相比,MSGCM模型可以有效支持多源地理空间信息的关联,进而能够支持地理空间信息查询、分析及综合展现等多种地理空间应用.实验及分析表明,MSGCM可以有效提高多源地理空间信息关联检索结果的多样性,并具备一定的可扩展性.

Geospatial information usually contains vector data, raster data and associated textdescription, which are interrelated with each other. How to retrieve and locate these correlatedinformation fast and comprehensively is a new demand for geospatial applications. In order toimprove the performance of retrieval and analysis of geospatial data, we proposed a Multi-SourceGeospatial data Correlation Model （MSGCM）. By extracting multiple features such as geo-location,semantic description, visual content and their mutual correlations, feature element graphs （FEG）are constructed first. Then, based on the predefined correlation schema, the multi-source geospa-tial objects are integrated into a unified space. Through calculating correlational strength of differentgeospatial objects, the graph-like correlation model is constructed. In order to improve the effi-ciency of model construction, we proposed a block-based optimization strategy with separatefeature indices. Compared with existing methods, MSGCM supports multi-source geospatial infor-mation correlation efficiently, and further can support several other geographic applications suchas geospatial information query, analysis and multi-view visualization. Experimental and analyticalresults show that MSGCM can enhance the diversity of query results with high scalability.