Visual Search and Detection in Background Clutter

Abstract

Two traditions of vision modeling have coexisted for many years with little or no transfer of information between them. Those interested in models of visual target acquisition for real-world scenarios have developed engineering models, which are essentially empirical summaries of visual performance data. On the other hand, basic researchers in visual psychophysics and neurophysiology have developed quantitative models of pattern perception. The basic research models have increased in generality and scope to the point that they are potentially powerful tools for addressing certain real-world needs that have recently come to the fore. The needs include quantitative, theory-based methods for evaluating target signatures, effects of background clutter, and observer false alarm rates. This paper reviews the shortcomings of existing target acquisition models, and reports work in progress to develop an improved model of target acquisition that incorporates a model of pattern perception from basic vision research.