Assessing Multispectral Image Fusion with Systems Factorial Technology

Abstract

Despite the rich literature on techniques for creating a single image from multispectral sensors, there is relatively little research on methods for assessing these techniques based on human performance. We propose the use of Systems Factorial Technology (SFT), a nonparametric, mathematical modeling framework for analyzing human cognition. Previous work has demonstrated the use of SFT in evaluating human perception of multi-spectral imagery, although on relatively contrived tasks. In this work, we extend the approach to a task in which observers must determine whether a person in the image is holding a gun or a tool. We found that all observers processed the information from each spectrum less efficiently when images based on two different spectra were presented together, regardless of whether the information was fused into a single image or kept separately. Furthermore, when images from the two different spectra were presented side-by-side, some observers were able to use both sources in parallel.