Enhancement of Component Images of Multichannel Data by Denoising with Reference

Abstract

Multichannel (multispectral, hyperspectral) remote sensing data may include junk component images which have quality considerably worse than other components. This can be due to intensive noise or low dynamic range of information component in such junk channels (sub-bands). Such component images are sometimes ignored (not used in further processing or analysis). Meanwhile, they can be subject to pre-filtering (denoising) in order to enhance them. To do this effectively, we propose to exploit the so-called reference images – component images that have relatively high quality and that are characterized by high similarity with respect to the image subject to pre-filtering. Within the framework of this general idea, we study several particular problems: how to choose component images that can be exploited as references, what transformations of reference images can be used, and how to perform the denoising. It is shown that one can employ as references component images of the same resolution as junk ones as well as component images of a better resolution. Not only one but also two references can be exploited. Different filters can be used after decorrelation. Enhancement of filtered images according to different quality criteria is provided. Examples of denoising for real-life multichannel images are given demonstrating high efficiency of the proposed approach.