Sj. Zhang et al., THE AUTOMATIC CONSTRUCTION OF A VIEW-INDEPENDENT RELATIONAL MODEL FOR3-D OBJECT RECOGNITION, IEEE transactions on pattern analysis and machine intelligence, 15(6), 1993, pp. 531-544
This paper describes and demonstrates a view-independent relational mo
del (VIRM) in a vision system designed for recognizing known 3-D objec
ts from single monochromatic images of unknown scenes. The aim is to e
stablish a model of an object, based on a CAD description, which is su
itable for its recognition without invoking pose information. We show
how the system can generate a VIRM automatically by a ''viewing and re
asoning'' process. The system inspects the CAD model from a number of
different viewpoints, and a statistical inference is applied to identi
fy relatively view-independent relationships among component parts of
the object. These relations are stored as a relational model of the ob
ject, which is represented in the form of a hypergraph. Three-dimensio
nal component parts (model features) of the object, which can be assoc
iated 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 hypergr
aph, and the pairwise view-independent relations form procedural const
raints associated with the hypergraph edges. During the recognition ph
ase, the covisibility measures allow a best-first search of the graph
for acceptable matches.