Beyond the Uniqueness Assumption: Ambiguity Representation and Redundancy Elimination in the Computation of a Covering Sample of Salient Contour Cycles

摘要

Perceptual organization provides an intermediate representation of data by means of object- and goal-independent information. The lack of complete information makes perceptual organization an intrinsically ambiguous process which invalidates the uniqueness assumption and requires instead the generation of multiple solutions. This raises the issue of eliminating redundancies which, in a recursive algorithm, might otherwise cause combinatorial explosion of the search space. These aspects of perceptual organization are illustrated in the context of cycle detection in a contour graph. A provably correct algorithm for this problem is proposed.