An evidential reasoning approach for recognizing shape features

This paper introduces an evidential reasoning based approachfor recognizing and extracting manufacturing featuresfrom solid model descriptionof objects. A major dificultyfaced by previouslyproposed methodsforfeature extraction has been the interaction betweenfeatures. In interacting situations, the representation for various primitivefeatures is non-unique,making their recognition very difficult. We develop an approach based on generating and combining geometric and topological evidencesfor recognizing interactingfeatures. The essence of our approach is in finding a set of correct and necessary virtual links throughaggregating the available geometric and topologic evidences at different abstraction levels. The identifed virtual links are then augmented to the cavity graphrepresenting a depression of an object so that the resulting supergraph can be partitioned to obtain the features of the object. The main contributions of our approach include introducing the evidential reasoning (Dempster-Shafer theory) to the feature extraction domain and developing the theory of principle of association to overcome the mutual exclusivenessassumption of the Dempster-Shafertheory.