Resolution enhancement of hyperspectral imagery through spatial-spectral data fusion

Abstract

Spatial resolution is one of the most expensive and hardest to improve in imaging systems. An efficient spatial-spectral data fusion method for superresolution of hyperspectral (HS) imagery through exploiting the spatial correlation of the endmembers using a superresolution mapping (SRM) technique is proposed in this paper. Endmember abundances (fractional images) obtained using linear mixture model and fully constrained least squares spectral unmixing algorithm, are processed using a spatial-spectral information correlation model and a learning-based SRM technique. The key element of the proposed technique is adopting a spatial-spectral correlation model through a learning based SRM algorithm to efficiently exploit the spatial dependence of the endmembers within any pixel of the HS data. The obtained results validate the reliability of the technique. The proposed method is independent of any high resolution secondary source of data and is low in terms of computational cost which makes it favorable for real-time applications.