A spatial relation ontology using mathematical morphology and description logics for spatial reasoning

In several domains, such as medical image interpretation, spatial relations play a crucial role since they are less prone to variability than properties of structures themselves. Moreover, they constitute an important part of available knowledge. In this paper, we suggest a new way to handle imprecise spatial information in a description logic framework based on mathematical morphology with fuzzy interpretations in concrete domains. Fuzzy representations of concepts contribute to bridge the gap between ontological representations and real data, and the mathematical morphology setting is a core feature of our approach to merge the advantages of ontologies, concrete domain representations and manipulations, and the reasoning power of description logics. The benefits of our logical framework are illustrated on a medical image interpretation task.