Computational models for search and discrimination

We present an experimental framework for evaluating metrics for the search and discrimination of a natural texture pattern from its background. Such m etrics could help identify preattentive cues and underlying models of searc h and discrimination, and evaluate and design camouflage patterns and autom atic target recognition systems. Human observers were asked to view image s timuli consisting of various target patterns embedded within various backgr ound patterns. These psychophysical experiments provided a quantitative bas is for comparison of human judgments to the computed values of target disti nctness metrics. Two different experimental methodologies were utilized. Th e first methodology consisted of paired comparisons of a set of stimuli con taining targets in a fixed location known to the observers. The observers w ere asked to judge the relative target distinctness for each pair of stimul i. The second methodology involved stimuli in which the targets were placed in random locations unknown to the observer. The observers were asked to s earch each image scene and identify suspected target locations. Using a pro totype eye tracking testbed, the integrated testbed for,eye movement studie s, the observers' fixation points during the experiment were recorded and a nalyzed. For both experiments, the level of correlation with the psychophys ical data was used as the basis for evaluating target distinctness metrics. Overall, of the set of target distinctness metrics considered, a metric ba sed on a model of image texture was the most strongly correlated with the p sychophysical data. (C) 2001 society of Photo-Optical Instrumentation Engin eers.