A Global Framework for Qualitative Shape Description

Qualitative models of spatial knowledge concern the description of both the shape of objects and their relative position in space. While there are contributions related to the latter aspect, small confidence is available for the former one. This paper aims to give a novel contribution in this direction. Qualitative models of shape should be seen as models able to represent commonsense knowledge of shape and not as a surrogate of quantitative models, which have a specific role in computer graphics and pattern recognition. Qualitative models are especially suitable for describing shapes in geographic space and build Geographical Information Systems that simulate the mental processes of human beings. The framework that we propose is structured in three dimensions (topological, projective, and metric properties of shape) that are shown to be necessary. Major emphasis is given to metric properties (elongation and symmetry) which have not received much attention till now.