Restructuring Aspect Graphs into Aspect- and Cell-Equivalence Classes for Use in Computer Vision

John H. Stewman, Louise Stark, Kevin W. Bowyer

Abstract: A potential disadvantage of using aspect graphs as object models in a computer vision system is their large size. The upper bound on the number of nodes in the aspect graph of an N-face convex object is O(N⋆⋆3). In this paper we introduce the concepts of "aspect-equivalence" and "cell-equivalence" and present a method of using them to restructure a data base of aspect graphs. This process identifies an interconnected set of equivalence classes which we use to form an Equivalence Class Graph (ECG).