This project is concerned with qualitative descriptions of geographic space and reasoning about spatio-temporal changes. Modeling time in geographic information systems has been investigated for a while and more recently, spatio-temporal reasoning has become of interest in the research community. Other recent approaches to space-time modeling have focused on query languages for the retrieval of video, e.g., through sketching. Our approach to modeling and reasoning about spatio-temporal change is based on a qualitative representation of spatial scenes, linking objects with identities through qualitative spatial relations. These relations will be based on a comprehensive model of topological spatial relations, as derived from the 4-intersection model and their metric refinements. From our work during the first two year on Heterogeneous Geographic Databases, where we investigated mechanisms to formalize consistency in multi-resolution geographic databases and to detect similarities among scenes of objects, we have obtained tools to reason about spatial relationships when objects change their structure, such as eliminating and merging holes or forming aggregates of disconnected parts. We also developed computational models to describe spatial similarity based on topological, directional, and metrical properties. We will build on these findings and apply them to developing qualitative descriptions of spatio-temporal change, analyzing sequences of spatial scenes for change, and predicting from a qualitative change description likely upcoming scenarios. We will focus on three specific aspects:
Last updated on July 1, 1998.
[ Geographic Databases | Spatial Database Research Group | NCGIA Maine ]