Coupled Tensor Block Term Decomposition with Superpixel-Based Graph Laplacian Regularization for Hyperspectral Super-Resolution

Hyperspectral image (HSI) super-resolution aims at improving the spatial resolution of HSI by fusing a high multispectral (MSI). To preserve local submanifold structures in super-resolution, novel superpixel graph-based method is proposed. Firstly, MSI segmented into blocks to form two-directional feature tensors, then two graphs are created using spectral–spatial distance between unfolded tensors. Secondly, graph Laplacian terms involving underlying BTD factors high-resolution developed, which ensures inheritance geometric structures. Finally, incorporating priors with coupled degradation model, model established. Experimental results demonstrate that proposed achieves better fused compared other advanced methods, especially on improvement structure.