Integral opponent-colors features for computing visual target distinctness

摘要

This paper presents a computational target distinctness model to predict human visual search performance for a set of color images. Visual distinctness metrics can be used to compare and rank target detectability, and to quantify background or scene complexity. The subjective ranking induced by the psychophysically determined visual target distinctness for 64 human observers is adopted as the reference order. The ranking produced by the proposed measure and several computational models are compared with the subjective rank order.