Evaluation of operator performance using true color and artificial color in natural scene perception

Abstract

Currently, the two most commonly used night optical devices employed in military operations are the long-wave infra-red sensor and the image intensified sensor. Recent advances in technology have permitted the fusion of the output of these two devices into a single color display that potentially combines the capabilities of both sensors while overcoming their limitations. Although the concept is appealing, previous sensor fusion studies have been inconclusive on the benefits of an artificially colored target. Perhaps, an artificially colored target disrupts an operator's visual processing thereby hindering the detection of a target. The purpose of this thesis is to compare the effects of artificial color, natural color, and monochrome formats in visual scene perception. It is hypothesized that participant response times and error rates would be greater at detecting an artificially colored target compared to a natural colored or a target presented achromatically. Two experiments were conducted. Experiment 1 used non-degraded imagery and Experiment 2 used degraded imagery to compare these effects. It was found that reaction time and error rates for naturally colored and achromatic images were similar and substantially less when compared to artificially colored images. For degraded scenes, natural color was more beneficial when compared to achromatic and artificially colored scenes. Additionally, artificially colored scenes caused extremely large error rates and reaction times. These results will provide algorithm developers insight into the importance of color constancy