Integrating gazetteers and remote sensed imagery

This work explores the potential for increased synergy between gazetteers and high-resolution remote sensed imagery. These two data sources are complementary. Gazetteers provide high-level semantic information about what is where but they must be manually compiled and maintained. On the other hand, imagery can be automatically acquired but only provides low-level radiometric information. We explore ways in which these two data sources can be integrated to more fully automate geographic data management. In particular, we show how gazetteers represent a rich source of semi-supervised training data for geospatial object modelling. We also describe an example of information flow in the other direction, namely, how high-resolution imagery can be used to refine the spatial extents of geospatial objects in gazetteers.