Bayesian Hyperspectral Image Super-Resolution in the Presence of Spectral Variability

Synthesizing a high-resolution (HR) hyperspectral image (HSI) by merging a low-resolution (LR) HSI with a corresponding HR multispectral image (MSI) has become a promising HSI super-resolution scheme. Most existing HSI-MSI fusion methods are effective to some extent, while several challenges remain. First, the spectral response of a given material exhibits considerable variability due to different acquisition times and conditions, however, variations in spectral signatures are often neglected. Second, a majority of off-the-shelf methods require predefined degradation operators, which can be unavailable in practice. To tackle the above issues, we introduce a novel fusion approach with a Bayesian framework. Specifically, we regard the up-sampled LR-HSI as the low-frequency component of the underlying HR-HSI. We characterize the texture features of high- and low-frequency components, respectively, which can enlarge modeling capacity and bypass the absence of degradation operators. Furthermore, we depict the relative smoothness of reflectance spectra with the Gaussian process. Extensive experiments on synthesized and real datasets illustrate the superiority of the proposed strategy in terms of fusion performance and robustness to spectral variability.

Automated summary

Synthesizing a high-resolution hyperspectral image by merging a low-resolution hyperspectral image with a corresponding high-resolution multispectral image has shown promise as a super-resolution scheme. However, challenges such as spectral variability and the lack of predefined degradation operators remain. This paper introduces a novel fusion approach with a Bayesian framework, which addresses these issues by characterizing texture features and using Gaussian processes.