Mixed Noise Reduction in Hyperspectral Imagery

In this paper, a hyperspectral mixed noise reduction technique is proposed called HyMiR. We assume that hyperspectral images include Gaussian and sparse noise types. First, the Gaussian noise is removed using a recently-developed method called hyperspectral restoration (HyRes). Then, the sparse noise is reduced by proposing a sparse noise removal technique. Experimental results confirm improvements in terms of signal to noise ratio and spectral angle distance. In addition, visual improvements can be clearly observed in the case of real dataset experiments.