PROJECT TITLE :

Nonlocal Patch Tensor Sparse Representation for Hyperspectral Image Super-Resolution

ABSTRACT:

Super-resolution of hyperspectral images (HSI) is shown in this study, which combines low-resolution (LR) HSI with high-resolution multispectral images (HR) to produce high-resolution HSI (HR-HSI). The nonlocal patch tensor is initially extracted from the nonlocal similar patches in order to build the proposed technique (NPT). To model the extracted NPTs, a new tensor sparse representation based on tensor-tensor product (t - product) is presented. To preserve both the spectral and spatial similarity of nonlocally identical patches, we use tensor sparse representation. In order to create a single objective function that incorporates nonlocal similarity, lexical learning, and sparse coding, the relationship between the HR-HSI and the LR-HSI is established using t - product. In the end, the optimization problem is solved using the alternating direction approach of multipliers. Three data sets and a real data set show that the suggested method significantly outperforms the current state-of-the art HSI super-resolution methods.