Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Deep unfolding Network for Hyperspectral Image Super-Resolution with Automatic Exposure Correction

Yuan Fang, Yipeng Liu, Jie Chen, Zhen Long, Ao Li, Chong-Yung Chi, Ce Zhu

TL;DR

This work addresses HSI-SR under heterogeneous exposure conditions by introducing UHSR-AEC, a deep unfolding network that jointly performs exposure correction and hyperspectral fusion. It combines a mathematically grounded exposure-aware degradation model with learnable proximal updates, an Initialization Module, and learned sampling operators to handle cross-sensor mismatches. The method demonstrates state-of-the-art performance on benchmark datasets, supported by ablation studies that confirm the importance of the IM and the unfolding depth. The approach offers a practical pathway to reliable HSI-SR in low-light or night-time scenarios where exposure mismatches are common, enhancing texture and spectral fidelity for downstream tasks.

Abstract

In recent years, the fusion of high spatial resolution multispectral image (HR-MSI) and low spatial resolution hyperspectral image (LR-HSI) has been recognized as an effective method for HSI super-resolution (HSI-SR). However, both HSI and MSI may be acquired under extreme conditions such as night or poorly illuminating scenarios, which may cause different exposure levels, thereby seriously downgrading the yielded HSISR. In contrast to most existing methods based on respective low-light enhancements (LLIE) of MSI and HSI followed by their fusion, a deep Unfolding HSI Super-Resolution with Automatic Exposure Correction (UHSR-AEC) is proposed, that can effectively generate a high-quality fused HSI-SR (in texture and features) even under very imbalanced exposures, thanks to the correlation between LLIE and HSI-SR taken into account. Extensive experiments are provided to demonstrate the state-of-the-art overall performance of the proposed UHSR-AEC, including comparison with some benchmark peer methods.

Deep unfolding Network for Hyperspectral Image Super-Resolution with Automatic Exposure Correction

TL;DR

This work addresses HSI-SR under heterogeneous exposure conditions by introducing UHSR-AEC, a deep unfolding network that jointly performs exposure correction and hyperspectral fusion. It combines a mathematically grounded exposure-aware degradation model with learnable proximal updates, an Initialization Module, and learned sampling operators to handle cross-sensor mismatches. The method demonstrates state-of-the-art performance on benchmark datasets, supported by ablation studies that confirm the importance of the IM and the unfolding depth. The approach offers a practical pathway to reliable HSI-SR in low-light or night-time scenarios where exposure mismatches are common, enhancing texture and spectral fidelity for downstream tasks.

Abstract

In recent years, the fusion of high spatial resolution multispectral image (HR-MSI) and low spatial resolution hyperspectral image (LR-HSI) has been recognized as an effective method for HSI super-resolution (HSI-SR). However, both HSI and MSI may be acquired under extreme conditions such as night or poorly illuminating scenarios, which may cause different exposure levels, thereby seriously downgrading the yielded HSISR. In contrast to most existing methods based on respective low-light enhancements (LLIE) of MSI and HSI followed by their fusion, a deep Unfolding HSI Super-Resolution with Automatic Exposure Correction (UHSR-AEC) is proposed, that can effectively generate a high-quality fused HSI-SR (in texture and features) even under very imbalanced exposures, thanks to the correlation between LLIE and HSI-SR taken into account. Extensive experiments are provided to demonstrate the state-of-the-art overall performance of the proposed UHSR-AEC, including comparison with some benchmark peer methods.
Paper Structure (14 sections, 7 equations, 4 figures, 2 tables)

This paper contains 14 sections, 7 equations, 4 figures, 2 tables.

Table of Contents

  1. Introduction
  2. NOTATIONS AND PROBLEM FORMULATION
  3. Notations
  4. HSI-SR Degradation Model
  5. PROPOSED METHOD
  6. Exposure Correction HSI Fusion Model
  7. Deep Unrolling on UHSR-AEC
  8. Initialization Module
  9. Loss Function
  10. Experiments
  11. Experimental settings
  12. Comparison with competitive methods
  13. Ablation studies
  14. Conclusion

Figures (4)

  • Figure 1: Block diagram of generating HR-HSI $\cal Z$ from the acquired HSI $\cal X$ and MSI $\cal Y$ under different exposure levels, for (a) most existing methods and (b) the proposed UHSR-AEC.
  • Figure 2: (a) The proposed UHSR-AEC, which consists of Initialization Module and Unfolding Module; (b) regularization step module, comprising two identical ResConvBlocks; (c) Initialization Module, comprising ConvBlock, Cross-attention, and Unet; (d) legend in the proposed UHSR-AEC structure.
  • Figure 3: Reconstructed images of various methods for the fake_and_real_peppers_ms (CAVE) dataset are illustrated by the false color image of [30,15,10] bands in Case 1.
  • Figure 4: Reconstructed images of various methods for the Imge3 (Harvard) dataset are illustrated by the false color image of [30,15,10] bands in Case 2.