Hyperspectral target detection from incoherent projections: Nonequiprobable targets and inhomogeneous SNR

This paper describes a computationally efficient approach for the detection of spectral targets of different strengths, contaminated by a colored Gaussian background, from relatively few incoherent projections compared to the dimension of the target. The performance of the detector is analyzed with respect to the number of observations collected, the background perturbation, the target signal strength, the similarity among different targets in a known spectral dictionary, and the known a priori probabilities of the targets.