MIT Department of Electrical Engineering & Computer Science

ML Detection in Hyperspectral Imagery

José M. F. Moura
Professor, Department of Electrical and Computer Engineering, Carnegie Mellon University

Tuesday, November 16, 1999
4:00 PM (reception following)
Room 35-225
LIDS Colloquium

Abstract

Hyperspectral sensors collect hundreds of narrow and contiguously spaced spectral bands of data (in the visible to long-wave infrared), organized in a hyperspectral cube. The hyperspectral imagery provides fully registered high resolution spatial and spectral information that is invaluable in discriminating between man-made objects and natural clutter backgrounds. This comes at a cost. The high volume of data in the hyperspectral cube has precluded the development of computationally practical Maximum-Likelihood (ML) detectors of man-made anomalies in clutter.

We address this problem. We derive the Gauss-Markov random field (GMrf) detector, a computationally efficient ML anomaly detector that fully adapts to the unknown statistics of the clutter and exploits the spatial and spectral correlation of the hyperspectral imagery. Test results with several hyperspectral sensors show significant improvement over other anomaly detectors. Due to its computational simplicity and good performance, the GMrf anomaly detector has been incorporated into the Adaptive Spectral Reconnaissance Program (ASRP) as one of the baseline processing algorithms.