Recently, we initiated research that attempts to describe performance of the subpixel detectors using analytical formulas. We showed a surprising result that MFD always outperformed OSP in a traditional theoretical formulation of the detection problem. We also showed that this theoretical formulation was not realistic in practical target detection in real spectral images. The purpose of this paper is to give specific formulas for the detection power of MFD and OSP detectors under a more realistic scenario when the background signatures vary in both target and non-target pizel spectra. In this paper, we consider normal (Gaussian) distributions of the abundances of the background signatures. We show under what conditions the MFD outperforms OSP, and when the opposite is true. We show three numerical examples of the balance between the bias of MFD and the large variability if OSP and its impact on the relative performance of these detectors.
Department, Program, or Center
The John D. Hromi Center for Quality and Applied Statistics (KGCOE)
©2005 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.
RIT – Main Campus