Hyperspectral signal models and implications to material detection algorithms

The paper presents a concise overview of hyperspectral signal models and the target detection algorithms resulting from their adoption. We focus on detection algorithms derived using established statistical techniques and whose performance is predictable under reasonable assumptions about hyperspectral imaging data. We show that the family of elliptically contoured distributions (ECDs), in general, and the t-ECD, in particular, provide a more accurate model for hyperspectral backgrounds, compared to the widely used multivariate normal distribution. Since many detection algorithms derived for normal distributions apply to ECDs as well, the ECD models provide a better framework for modeling and analyzing hyperspectral imaging data.