Efficient Feedback Gate Network for Hyperspectral Image Super-Resolution

2025-06-20

AI Summary

To address insufficient spectral band coherence modeling and inadequate spatial-spectral feature fusion in single-image hyperspectral super-resolution (SHSR), this paper proposes a novel joint spatial-spectral modeling framework. Methodologically, it introduces three key innovations: (1) a group-wise feedback gating mechanism for dynamic inter-band dependency learning; (2) the Spectral-Progressive Dilated Fusion Module (SPDFM), integrating band-guided channel reordering with progressive dilated convolutions to enhance multi-scale spatial-spectral representations; and (3) the 3D Spatial-Spectral Residual Gating Module (SSRGM), jointly optimizing global spatial-spectral consistency via wide-receptive-field perception and spectral enhancement gating. The framework incorporates large-kernel convolutions, channel shuffling, and a 3D spatial-spectral gating structure. Extensive experiments on three benchmark datasets demonstrate state-of-the-art performance, achieving superior spectral fidelity (lowest SAM) and spatial reconstruction quality (highest PSNR/SSIM).

Technology Category

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

Application Category

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

Abstract

Even without auxiliary images, single hyperspectral image super-resolution (SHSR) methods can be designed to improve the spatial resolution of hyperspectral images. However, failing to explore coherence thoroughly along bands and spatial-spectral information leads to the limited performance of the SHSR. In this study, we propose a novel group-based SHSR method termed the efficient feedback gate network, which uses various feedbacks and gate operations involving large kernel convolutions and spectral interactions. In particular, by providing different guidance for neighboring groups, we can learn rich band information and hierarchical hyperspectral spatial information using channel shuffling and dilatation convolution in shuffled and progressive dilated fusion module(SPDFM). Moreover, we develop a wide-bound perception gate block and a spectrum enhancement gate block to construct the spatial-spectral reinforcement gate module (SSRGM) and obtain highly representative spatial-spectral features efficiently. Additionally, we apply a three-dimensional SSRGM to enhance holistic information and coherence for hyperspectral data. The experimental results on three hyperspectral datasets demonstrate the superior performance of the proposed network over the state-of-the-art methods in terms of spectral fidelity and spatial content reconstruction.

Problem

• Improving spatial resolution of single hyperspectral images without auxiliary data
• Enhancing band coherence and spatial-spectral information exploration
• Achieving superior spectral fidelity and spatial content reconstruction

Innovation

• Feedback gate network with large kernel convolutions
• Shuffled progressive dilated fusion module
• Spatial-spectral reinforcement gate module