Abstract
Several types of sensors are used to capture imagery to provide information to human observers. Images from each sensor can be presented to an observer individually, or the images from multiple sensors can be algorithmically combined into a single image. The various algorithms for fusing images have been studied from an information theoretic viewpoint, but the advantages (or disadvantages) of fusion for human perceptual and cognitive processing have received relatively little attention. Systems Factorial Technology (SFT; Townsend & Nozawa, 1995) is general framework for assessing of how human observers process multiple sources of information. We applied SFT to measure how observers perform with images from different sensors separately (side-by-side). This performance was used as a baseline to compare performance with fused imagery. Because there are two separate images in the side-by-side presentation, there may be some statistical facilitation of the processing times, while fused presentations eliminate the need to attend to both sides. These two potential gains seem to trade off we found roughly equivalent workload capacity levels, limited for all participants, in both conditions. The survivor interaction contrast (SIC), another measure of SFT, indicated most individuals processed the side-by-side images sources of information simultaneously (in parallel) while responding with the first completed source. Despite the roughly equal capacity, the fused images result in slightly faster response times than the redundant side by side images at the group level, with all sensor types being equally fast with no significant interaction. While the fused imagery is processed slightly faster, performance with the side-by-side presentation is quite good because participants were able to process the images in parallel. Given that there is necessarily some loss of information in the fused images, it may be use side-by-side images even in time critical applications.
Meeting abstract presented at VSS 2014