Global Feature Schemes for Qualitative Shape Descriptions

Abstract

Qualitative shape descriptions are important in a number of fields in AI, such as in qualitative reasoning, and especially in robotics and vision. Current approaches are confined to describing local features while walking along the shapes’ boundaries. Shapes exist, however, which cannot be distinguished by these methods even though there are obvious distinctions between them. We shall introduce the notion of a global feature scheme in order to compensate for the shortcomings of present techniques. This approach is then applied to a class of shapes which have previously been shown to be difficult to distinguish, and proves capable of telling them apart. The method is based on a representation of intersection-free relations which complements existing qualitative representations for which intersections are virtually fundamental.1