THE INFORMATIONAL DIFFERENCE CONCEPT IN ANALYZING TARGET RECOGNITION ISSUES

A model for analyzing issues involving monospectral target recognition is presented. These issues include modeling target detection, recogni tion and identification thresholds, and predicting the functional para metric dependencies of the results of observation experiments with hum an observers. The model makes extensive use of concepts used in inform ation theory. An image of a scene is treated as a sample of an entire set of images of that scene. A difference measure, called the informat ional difference (InDif) between two image sets is defined. The main a ssertion is that accomplishing target recognition tasks is equivalent to setting thresholds for the InDif. The applicability of the InDif to the performance of the human visual system (HVS) is shown both analyt ically, in very simple situations, and in computer calculations involv ing noisy images. Finally, a single framework for dealing with the HVS and artificial intelligence systems in target recognition application s is shown to result naturally from the InDif formalism. (C) 1997 Opti cal Society of America.