GeoMatchMaker: Automatic and Efficient Matching of Vector Data with Spatial Attributes in Unknown Geometry Systems

Large amount of geospatial data are now available from public and private organizations in vector data formats. Users of these geospatial data usually require the data that are gathered from different sources to be integrated and fused for knowledge discovery. A vital step for fusion of the geospatial datasets is to identify the matched features among the datasets. There have been several efforts to automatically or semi-automatically detect matched features across different vector datasets. These solutions usually require the features to be in the same coordinate system so their spatial attributes can be compared. This renders these solutions impractical for the scenarios where the coordinate systems of the datasets are unknown. In this paper we propose several approaches that are based on utilizing the intersections of the lines as features, to efficiently and accurately detect the matched features across line vector datasets. We first discuss PPM, a brute-force approach to find the transformation of the intersections from one dataset to another. We then briefly discuss Geo-PPM, an improvement over PPM that utilizes some network properties to prune the search space. Finally, we discuss prioritized Geo-PPM where we can further improve the performance of Geo-PPM by first examining the features that have a higher possibility of locating the matching pattern. Our experiments show that prioritized GeoPPM provides a substantial improvement over Geo-PPM and hence, renders Geo-PPM practical for networks with large sizes.