Class-based recognition of 3D objects represented by volumetric primitives

Abstract

This paper presents a novel approach to recognizing 3D complex objects that have similar geometric structure but belong to different subclasses. Test scenes are acquired by a laser striper as range images and the objects are modelled using a composite volumetric representation of superquadrics and geons. Matching is decomposed into two stages: first, an indexing scheme designed to make effective use of the symbolic keys of the representation is implemented in order to direct the search to the models containing the parts identified; second, a method is proposed where the hypotheses picked from the index are searched using an Interpretation Tree algorithm combined with a quality measure to evaluate the bindings and the final valid hypotheses based on Possibility Theory, or the Theory of Fuzzy Sets. The valid hypotheses ranked by the matching process are then passed to the pose estimation module. From the results shown here, we can conclude that shape and pose parameter estimation from range data is sufficiently accurate to allow the algorithms used here to discriminate between objects having the same general topology, but different specific shapes.