COMPUTATIONAL VISUAL DISTINCTNESS METRIC

A new computational visual distinctness metric based on principles of the early human visual system is presented. The metric is applied to q uantify (1) the visual distinctness of targets in complex natural scen es and (2) the perceptual differences between compressed and uncompres sed images. The new metric is shown (1) to predict human observer perf ormance in search and detection tasks on complex natural imagery, (2) to correlate with visual target distinctness estimated by human observ ers, and (3) to correlate with the quality factor of Joint Photographi c Experts Group (JPEG) compressed imagery. (C) 1998 society of Photo-O ptical Instrumentation Engineers.