The Locus model of search and its use in image interpretation

Abstract

The Locus model of search is a non-backtracking, deterministic search technique in which a beam of near-miss alternatives around the best path are extended in parallel for graph searching problems. In this paper we formulate image interpretation as a giaph searching problem and show bow the Locus model provides a near-optimal minimal effort solution. The structure of the model is illustrated using a detailed example. The relationship of the present approach to earlier attempts at image interpretation are discussed.