A Framework for Integrating and Reasoning about Geospatial Data

The amount of geospatial data continues to proliferate with recent advancements in information technology. Available geospatial data sources include mapping services (GoogleMaps, YahooMaps, etc.), Web2.0 based collaborative projects (OpenStreetMaps and WikiMapia), traditional geospatial data sources (raster maps, KML vector layers, etc.) and non-traditional geospatial data sources (phonebooks and property-records). To fully exploit these diverse geospatial data sources, we are developing an integrated approach to extraction and fusion of these sources within a unified framework. A geospatial fusion framework needs to support both the integration and reasoning of heterogeneous geospatial data. The data integration tasks involve gathering the available geospatial data from a wide variety of sources, such as those listed above. The geospatial reasoning processes can infer new and useful knowledge about a region by applying various reasoning methods over the integrated data. Some common geospatial reasoning processes are creating 3D models of buildings from LIDAR (Zhou & Neumann, 2008), identifying streets from raster maps (Chiang et al., 2008), automatic conflating road vector data with orthoimagery (Chen et al., 2007), and so on. Figure 1 shows an example screenshot where a variety of data sources and reasoning capabilities have been integrated into a single integrated framework. In this figure, the fusion of the datasets and reasoning processing make it possible to identify the locations of the buildings, the names of the streets, and the businesses associated with each of the buildings.