USING PERCEPTUAL INFERENCE NETWORKS TO MANAGE VISION PROCESSES

We provide a probabilistic framework, based on Perceptual Inference Ne tworks, for the management of computational resources such as special purpose modules, feature detectors, and highly domain dependent algori thms. Since these resources tend to be: computationally expensive and have limited applicability, judicious management is warranted. The res ources are used to build a comprehensive description of the scene. Res ources are selected in an information theoretic framework with the max imization of information gain per unit of computation as the optimalit y criterion. The viability of the algorithm is demonstrated in percept ual organization tasks. (C) 1995 Academic Press, Inc.