Minimum error gain for predicting visual target distinctness

We present a new method for characterizing information about a target relative to its background. The resultant computational measures are then applied to quantify the visual distinctness of targets in complex natural backgrounds from digital imagery. A generalization of the Kullback-Leibler joint information gain over the optimal interest points of the target image is shown to correlate strongly with visual target distinctness as estimated by human observers. Optimal interest points are defined as spatial locations of partially invariant features, which minimize the error probability between the target and the nontarget scenes; their significance is a function of the corresponding degree of congruence across scales and orientations. (C) 2001 Society of Photo-Optical Instrumentation Engineers.