Hyperspectral-Multispectral Image Fusion via Tensor Ring and Subspace Decompositions
Honghui Xu, Mengjie Qin, Shengyong Chen, Yuhui Zheng, Jian wei Zheng
Abstract
Fusion from a spatially low resolution hyperspectral image (LR-HSI) and a spectrally low resolution multispectral image (HR-MSI) to produce a high spatial-spectral HSI (HR-HSI) has risen to a preferred topic for reinforceing the spatial-spectral resolution of HSI in recent years, that is additionally known as Hyperspectral super-resolution. In this work, we propose a new model, namely low-rank tensor ring decomposition based on tensor nuclear norm (LRTRTNN), for HSI-MSI fusion. Specifically, we reconstruct HR-HSIs via a novel tensor-ring (TR) approximation bonded with the tensor nuclear norm (TNN) constraint for better fusion results. All 3D tensor ring factors are no longer unfolded to suit the matrix nuclear norm used in conventional methods, while retaining their tensor structure and directly applying TNN to faithfully reflect the internal structure of the tensors. Moreover, the global low-rank subspace of LR-HSIs and the non-local similarity deriving from the patch-based structuralization of HR-MSIs are jointly considered to faithfully capture the inherent details of HR-HSIs compromised by slightly more computational cost. The Alternating Direction Method of Multipliers (ADMM) is introduced for coefficients optimization. Numerical and visual experiments on real data show that our LRTRTNN method outperforms a number of state-of-the-art algorithms in terms of fusing performance.