Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Hybrid Convolutional and Attention Network for Hyperspectral Image Denoising

Shuai Hu, Feng Gao, Xiaowei Zhou, Junyu Dong, Qian Du

TL;DR

HCANet tackles hyperspectral image denoising by jointly modeling local and global information through a convolution-attention hybrid. It introduces a convolution-attention fusion module (CAFM) and a multi-scale feed-forward network (MSFN) within a U-shaped CAMixing framework to enhance denoising across spatial, spectral, and multi-scale contexts. The approach is trained with a gradient-regularized reconstruction objective and demonstrates superior performance over state-of-the-art methods on benchmark HSIs under various noise conditions, while maintaining a compact model size. The work provides open-source code to facilitate reproducibility and further research in HSI denoising.

Abstract

Hyperspectral image (HSI) denoising is critical for the effective analysis and interpretation of hyperspectral data. However, simultaneously modeling global and local features is rarely explored to enhance HSI denoising. In this letter, we propose a hybrid convolution and attention network (HCANet), which leverages both the strengths of convolution neural networks (CNNs) and Transformers. To enhance the modeling of both global and local features, we have devised a convolution and attention fusion module aimed at capturing long-range dependencies and neighborhood spectral correlations. Furthermore, to improve multi-scale information aggregation, we design a multi-scale feed-forward network to enhance denoising performance by extracting features at different scales. Experimental results on mainstream HSI datasets demonstrate the rationality and effectiveness of the proposed HCANet. The proposed model is effective in removing various types of complex noise. Our codes are available at \url{https://github.com/summitgao/HCANet}.

Hybrid Convolutional and Attention Network for Hyperspectral Image Denoising

TL;DR

HCANet tackles hyperspectral image denoising by jointly modeling local and global information through a convolution-attention hybrid. It introduces a convolution-attention fusion module (CAFM) and a multi-scale feed-forward network (MSFN) within a U-shaped CAMixing framework to enhance denoising across spatial, spectral, and multi-scale contexts. The approach is trained with a gradient-regularized reconstruction objective and demonstrates superior performance over state-of-the-art methods on benchmark HSIs under various noise conditions, while maintaining a compact model size. The work provides open-source code to facilitate reproducibility and further research in HSI denoising.

Abstract

Hyperspectral image (HSI) denoising is critical for the effective analysis and interpretation of hyperspectral data. However, simultaneously modeling global and local features is rarely explored to enhance HSI denoising. In this letter, we propose a hybrid convolution and attention network (HCANet), which leverages both the strengths of convolution neural networks (CNNs) and Transformers. To enhance the modeling of both global and local features, we have devised a convolution and attention fusion module aimed at capturing long-range dependencies and neighborhood spectral correlations. Furthermore, to improve multi-scale information aggregation, we design a multi-scale feed-forward network to enhance denoising performance by extracting features at different scales. Experimental results on mainstream HSI datasets demonstrate the rationality and effectiveness of the proposed HCANet. The proposed model is effective in removing various types of complex noise. Our codes are available at \url{https://github.com/summitgao/HCANet}.
Paper Structure (9 sections, 9 equations, 5 figures, 4 tables)

This paper contains 9 sections, 9 equations, 5 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Methodology
  3. Convolution and Attention Fusion Module
  4. Multi-Scale Feed-Forward Network
  5. Loss Function
  6. Experimental Results and Analysis
  7. Experiment Setup
  8. Experimental Analysis
  9. Conclusion

Figures (5)

  • Figure 1: Illustration of our proposed hybrid convolution and attention network (HCANet) for HSI denoising. (a) Framework of HCANet. (b) Inner structure of CAMixing block.
  • Figure 2: Illustration of the proposed convolution and attention fusion module (CAFM). It consists of local and global branches. In the local branch, convolution and channel shuffling are employed for local feature extraction. In the global branch, the attention mechanism is used to model long-range feature dependencies.
  • Figure 3: Illustration of the multi-scale feed-forward network (MSFN).
  • Figure 4: Gaussian noise removal results under noise level $\sigma=50$ on ICVL dataset with bands (17,20,30).
  • Figure 5: Real noise removal results on Pavia dataset with bands (17, 20, 30).