QUALITATIVE SPATIAL REASONING UNDER UNCERTAINTY IN GEOlGRAPHICAL INFORMATION SYSTEMS

The capability of geographical information systems (GIS) to display images and analyse them is one of the most powerful, because people lnave a very consistent ability to reason about visual features according to their own knowledge model. As images are accepted as a symbolic view of the world, with familiar and intuitive entities, the set of tasks we desire for the performance Cc the GIS includes image understanding and image manipulation. Usually the abilities of GIS are restricted dealing with quantitative data only, so that they fail whenever exact matches can't be found. A solution for this problem is not only to allow quantitative information, but also qualitative one. Proximity factors such as 'close to' and 'far from' . The qualitative approach offers a more natural way for the user of the system to specify what he actually wants, by using proximity operators. This paper is an attempt of integrating qualitative spatial reasoning into geographic information systems. The idea is to associate qualitative information with fuzzy sets and to use the values of these fizzy sets fra spatial reasoning. The authors are concerned with the problem of image interpretation, which stands for the geometric and semantic recognition of the objects contained in one image.