For cases when the background clutter model and/or the target signal are not known with perfect fidelity, the standard matched filter can fail. Robust matched filters apply some regularization to the problem to develop a more reliable result. This task deals with this implementation.

References:

Manolakis, D., Lockwood, R., Cooley, T., & Jacobson, J. (2009, August). Hyperspectral detection algorithms: Use covariances or subspaces?. In Imaging Spectrometry XIV (Vol. 7457, p. 74570Q). International Society for Optics and Photonics.