ABSTRACT

This chapter describes how super-resolution techniques can be applied to improve both spatial and spectral resolution of the multichannel images. It outlines the super-resolution problem in a reduced dimensional spectral subspace mainly for hyperspectral data. The chapter introduces an application of super-resolution techniques to multichannel data, specifically to hyperspectral images. It shows that the requirement to design expensive equipment to capture high-resolution spectral data can be circumvented by fast, simple, and cost-efficient postprocessing methods. In the design of an imaging model for the super-resolution reconstruction, since the desired resolution level is higher than the resolution of the observed images, the blurring and the downsampling operations are usually incorporated into the model to explain the relation between the targeted and observed resolution. The advantage of using the global projective mapping for the motion compensation of low-resolution observations is because it enables a very accurate estimation of individual pixel locations at the corresponding high-resolution grids.