The Psychophysics of Sensor Fusion: A Multidimensional Signal Detection Analysis

Abstract

Image fusion techniques take input from multiple single-band sensors and combine it to create a single multi-band image. While such processing offers to create imagery that is more information rich than that produced by single-band sensors, it may sometimes degrade the perceptual quality of the input content. Working within the context of Ashby and Townsend's (1986) General Recognition Theory, a multidimensional signal detection model of perceptual interactions between stimulus features, the current study measured the perceptibility of single-band content within fused images. Data indicate that the perceptibility of information from one single-band input channel can be degraded as the contrast level of the alternative single-band input is manipulated. False-color rendering of fused images, likewise, can sometimes improve and sometimes degrade perceptibility of single-band content. Implications for the design and testing of sensor-fused displays are discussed.