14 July 1988 Unbiased Rigorous Figures Of Merit For Comparing LWIR Sensor Observables And/Or Observable Estimates
Author Affiliations +
Abstract
Figures of merit for the unbiased comparison of in-band radiant intensities or irradiances to model predictions or observations of the identical fluxes by a second sensor have been developed. For Gaussian errors the figures of merit obey x2 statistics for the appropriate degrees of freedom. Noise sources rigorously treated are: 1) time-band dependent signal-to-noise (S/N), 2) in-band precision (σ), 3) uncorrelated bias or band-to-band error (ε), and 4) correlated bias (A). The figures of merit are simple to apply because all required matrix inversions have been performed analytically. The figures of merit are based on a group theoretical-like decomposition of the Mahalanobis distance where the features are radiant intensities and sensoronly uncertainty contributes to the covariance matrix. The principal application of these formulae is expected to be model validation, although past related applications have been various.
© (1988) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
J. V. Beaupre, J. V. Beaupre, } "Unbiased Rigorous Figures Of Merit For Comparing LWIR Sensor Observables And/Or Observable Estimates", Proc. SPIE 0940, Infrared Scene Simulation: Systems, Requirements, Calibration, Devices, and Modeling, (14 July 1988); doi: 10.1117/12.947163; https://doi.org/10.1117/12.947163
PROCEEDINGS
10 PAGES


SHARE
Back to Top