MSDH: Matched subspace detector with heterogeneous noise

Abstract The matched subspace detector (MSD) is a classical subspace-based method for hyperspectral subpixel target detection. However, the model assumes that noise has the same variance over different bands, which is usually unrealistic in practice. In this letter, we relax the equal variance assumption and propose a matched subspace detector with heterogeneous noise (MSDH). In essence, the noise variances are different for different bands and they can be estimated by using iteratively reweighted least squares methods. Experiments on two benchmark real hyperspectral datasets demonstrate the superiority of MSDH over MSD for subpixel target detection.