Establishment of Human Performance Baseline for Image Fusion Algorithms in the LWIR and SWIR Spectra

Volume Number:
Issue Number:
Starting page
Ending page
Publication Date:
Publication Date
1 December 2013
Christopher Howell, Steve Moyer

paper Menu


This research is complementary to research presented in “Establishment of Human Performance Baseline for Image Fusion Algorithms in the LWIR and MWIR Spectra” by Moyer and Howell in which we established a baseline performance candidate for image fusion comparison by investigating the impact of different display formats on the probability of identification, P(ID), performance of a human observer. We advance this line of research by measuring the inherent ability of the human observer to perform an identification task using source band imagery, long-wave (LW) infrared and short-wave (SW) infrared that was not algorithmically fused prior to hu-man observation. A multi-part experiment was conducted where human observers were asked to identify displayed military targets using a standard set of tracked military vehicles. The observers performed the identification (ID) visual discrimination task using source band imagery concatenated or presented side-by-side on a single monitor, temporally interlaced source band imagery on a single monitor. Observers’ performances using source band imagery fused with the superposition fusion algorithm was also included as a reference because it is a well understood algorithm and shares an assumed similarity with the temporal interlaced display format. This research proposes that a forced choice human perception experiment is a more meaningful evaluation of an image fusion algorithm’s performance, specifically when the application of the algorithm is for dimensionality reduction to a single image designed for human observation. The results from this research identify a clear performance baseline when assessing human observer P(ID) performance employing an image fusion algorithm.