HSSDCT: Factorized Spatial-Spectral Correlation for Hyperspectral Image Fusion

2026-01-31

AI Summary

This work addresses key challenges in hyperspectral image fusion—limited receptive fields, spectral redundancy, and the high computational complexity of self-attention—which hinder reconstruction efficiency and robustness. To overcome these limitations, we propose the Hierarchical Spatial-Spectral Dense Correlation Transformer (HSSDCT), which incorporates Hierarchical Dense Residual Transformer Blocks (HDRTBs) to effectively enlarge the receptive field. Furthermore, a Spatial-Spectral Correlation Layer (SSCL) is introduced to explicitly decouple spatial and spectral dependencies, reducing the self-attention complexity to linear scale. By integrating window-progressive expansion with dense residual connections, the proposed method achieves state-of-the-art performance across multiple benchmark datasets, delivering superior reconstruction quality while significantly lowering computational costs.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Feature Construction/Reformulation
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Hyperspectral image (HSI) fusion aims to reconstruct a high-resolution HSI (HR-HSI) by combining the rich spectral information of a low-resolution HSI (LR-HSI) with the fine spatial details of a high-resolution multispectral image (HR-MSI). Although recent deep learning methods have achieved notable progress, they still suffer from limited receptive fields, redundant spectral bands, and the quadratic complexity of self-attention, which restrict both efficiency and robustness. To overcome these challenges, we propose the Hierarchical Spatial-Spectral Dense Correlation Network (HSSDCT). The framework introduces two key modules: (i) a Hierarchical Dense-Residue Transformer Block (HDRTB) that progressively enlarges windows and employs dense-residue connections for multi-scale feature aggregation, and (ii) a Spatial-Spectral Correlation Layer (SSCL) that explicitly factorizes spatial and spectral dependencies, reducing self-attention to linear complexity while mitigating spectral redundancy. Extensive experiments on benchmark datasets demonstrate that HSSDCT delivers superior reconstruction quality with significantly lower computational costs, achieving new state-of-the-art performance in HSI fusion. Our code is available at https://github.com/jemmyleee/HSSDCT.

Problem

• Hyperspectral Image Fusion
• Spatial-Spectral Correlation
• Spectral Redundancy
• Computational Complexity
• High-Resolution Reconstruction

Innovation

• Hyperspectral Image Fusion
• Spatial-Spectral Correlation
• Linear Complexity Self-Attention
• Hierarchical Dense-Residue Transformer
• Factorized Attention