Assessing the Consistency of Complete and Incomplete Topological Information

This work was partially funded by grants from Intergraph Corporation. Jayant Sharma was supported by a University Graduate Research Assistantship (UGRA) from the University of Maine. Additional support from NSF for the NCGIA under grant number SES 88-10917 is gratefully acknowledged. High-level topological information about spatial objects can be described in terms of a set of binary topological relations between the objects, also called a scene description. The objects of interest are , which are bounded objects that have a distinct identity and are homeomorphic to a 2-disk. The consistent integration of topological information relies inherently on the algebraic properties of the relations between the objects. Properties such as the converseness of pairs of relations and the composition of relations must be fulfilled for any combination of relations in order to guarantee that a scene description is free of internal topological contradictions so that it can be realized in IR . A rigorous computational method has been designed to reason about binary topological relations between spatial regions and to infer the consistency of complete and incomplete topological information. As a side-product, the method can be also used to refine incomplete observations. The method applies immediately in spatial query processing in geographic information systems to detect unsolvable queries prior to query execution and in data fusion to integrate independently collected information.