Image metrics for human search and discrimination of textured targets and backgrounds

This dissertation is a study of the use of quantitative image metrics to measure the signature strength of a camouflaged target in a sensed image. These measurements of signature strength are compared to the results of experiments in which human observers perform the tasks of search and discrimination with digital imagery. The real-world analogy of these experiments is that of a military observer looking upon the sensed image of a static scene to find camouflaged enemy targets that are re-ported to be in the area. Several pure discrimination experiments are conducted with image stimuli possessing target patterns in a fixed location. The psychological scale values from human judgments of the stimuli are compared to both existing and newly developed target signature metrics. Using this technique, it is demon-strated that some newly developed texture-based metrics most effectively quantify the perceptual distinctness of the target patterns. The task of search is integrated with the task of discrimination in an experiment in which observers are asked to view image stimuli containing targets randomly distributed throughout real-world outdoor scenes. The observer's fixation point in the image is measured by tracking his eye movements during the experiment. From the data, various statistics are computed for each target to describe how easily the observers located it, including the likelihood the target was fixated or identified and the time required to do so. The target signature metrics are compared to these statistics, and the texture-based metrics are again found to be the most highly correlated with the data.