SSDCN: Spatial-Spectral Dual-Clustering-based Network for Hyperspectral Image Super-resolution

Hyperspectral Image Single Image Super-Resolution (HSI-SISR) faces a conflict between computational efficiency and global non-local modeling. Existing Transformers suffer from quadratic complexity, while window-based methods compromise global capture. To address this, we propose the Spatial-Spectral Dual-Clustering-based Network (SSDCN). Our method introduces three innovations. First, we design a Spatial-Spectral Dual-Cluster Block (SSDCB). Replacing expensive point-to-point attention, it uses content-driven clustering to learn low-rank structural bases, achieving global modeling with linear complexity $\mathcal{O}(KN)$. Second, we propose a pyramid progressive hierarchical architecture with a Feature Reuse Reconstruction Block (FRRB). It reuses the core tensor and spectral factors from coarse levels, updating only spatial factors to minimize redundancy. Third, we propose a Pyramid Hierarchical Reconstruction Joint Loss to supervise intermediate levels, ensuring structural accuracy and preventing error accumulation. Experiments demonstrate that SSDCN surpasses SOTA methods in metrics and visual quality with significantly fewer parameters and FLOPs, achieving an optimal efficiency-performance balance.

Hyperspectral images capture a massive amount of light information far beyond what the human eye can see, but enhancing their resolution is highly challenging. Current AI models face a dilemma: they either analyze the entire image at a huge computational cost, or they save computer power but miss the "big picture." To solve this, we introduce a new AI method that efficiently balances both. Instead of comparing every single pixel to every other pixel—which is slow and memory-intensive—our approach cleverly groups similar visual and color patterns together. We also designed the system to build the high-resolution image step-by-step, recycling useful information along the way to avoid doing the same work twice. Furthermore, we implemented a checking mechanism at each step to prevent small mistakes from turning into large errors. As a result, our method produces sharper and more accurate images than current leading technologies, while using significantly less computational power.