THE AUTOMATIC CONSTRUCTION OF A VIEW-INDEPENDENT RELATIONAL MODEL FOR 3-D OBJECT RECOGNITION

This paper describes and demonstrates a view-independent relational model (VIRM) in a vision system designed for recognizing known 3-D objects from single monochromatic images of unknown scenes. The aim is to establish a model of an object, based on a CAD description, which is suitable for its recognition without invoking pose information. We show how the system can generate a VIRM automatically by a ''viewing and reasoning'' process. The system inspects the CAD model from a number of different viewpoints, and a statistical inference is applied to identify relatively view-independent relationships among component parts of the object. These relations are stored as a relational model of the object, which is represented in the form of a hypergraph. Three-dimensional component parts (model features) of the object, which can be associated with extended image features obtained by grouping of primitive 2-D features, are represented as nodes of the hypergraph. Covisibility of model features is represented by means of hyperedges of the hypergraph, and the pairwise view-independent relations form procedural constraints associated with the hypergraph edges. During the recognition phase, the covisibility measures allow a best-first search of the graph for acceptable matches.