Hyperspectral impulse denoising with sparse and low-rank penalties

Abstract

There are various studies on hyperspectral image denoising most of which consider Gaussian denoising problem. There are few studies on reducing impulse noise from correlated hyperspectral images. To reduce impulse noise, in our prior work we exploited the inter-band spectral correlation along with intra-band spatial redundancy to yield a sparse representation in transform domains. In this work, we improve upon the prior technique. The intra-band spatial redundancy is modeled as a sparse set of transform coefficients and the inter-band spectral correlation is modeled as a rank deficient matrix. The resulting optimization problem is solved using the Split Bregman technique. Comparative experimental results show that our proposed approach is better than the previous one in terms of visual quality.