In the first year of this project we developed tools for accurately and automatically integrating diverse geospatial data by exploiting what is known or can be inferred about each of the data sources. For the imagery, we exploit image metadata (such as ground resolution and geo-coordinates) as well as the color of imagery. For vector data of roads, we exploit metadata about the vectors, such as address ranges, road names, or even the number of lanes and type of road surface. For raster maps, we can exploit map scales, map geo-coordinates or perform image processing techniques (such as edge-detection) at the pixel level to extract primitive entities, such as line intersections. In this year, in particular, we focused on the integration of high resolution color imagery with various accuracy levels vector data. Moreover, we extended our vector-imagery fusion techniques to integrate multiple online maps with imagery. Furthermore, there has been relatively little work on automatically conflating maps with imagery. In , the authors describe how an edge detection process can be used to determine a set of features that can be used to conflate two image data sets. However, their work requires that the coordinates of both image data sets be known in advance. Dare and Dowman  proposed a feature-based registration technique to integrate two images. However, their approach requires users to manually select some initial control points. Some commercial GIS products, such as Able R2V14 and Intergraph I/RASC15 provide the functionality of conflating imagery and maps (i.e., raster to raster registration) using different types of transformation methods. However, these products do not provide automatic conflation, so users need to manually pick control points for conflation. Source.