Dual-domain aggregation Transformer for hyperspectral image super-resolution

WANG Haijun; WANG Yaowei; YANG Tengfei; HUO Limei; LI Peiluan

doi:10.16152/j.cnki.xdxbzr.2026-03-008

您当前的位置：

首页 >

文章列表页 >

Dual-domain aggregation Transformer for hyperspectral image super-resolution

更新时间：2026-07-07

- Dual-domain aggregation Transformer for hyperspectral image super-resolution
- Journal of Northwest University (Natural Science Edition) Vol. 56, Issue 3, Pages: 551-560(2026)
- 作者机构：
  
  1.河南科技大学　数学与统计学院，河南　洛阳　471023
  2.中国科学院　空天信息创新研究院，北京　100094
- 作者简介：
- 基金信息：
- DOI：10.16152/j.cnki.xdxbzr.2026-03-008
  CLC： TP391
- Received：20 January 2026，
  
  Revised：2026-01-30，
  
  Published：25 June 2026
- 稿件说明：
移动端阅览
WANG Haijun, WANG Yaowei, YANG Tengfei, et al. Dual-domain aggregation Transformer for hyperspectral image super-resolution[J]. Journal of Northwest University (Natural Science Edition), 2026, 56(3): 551-560.
DOI：

WANG Haijun, WANG Yaowei, YANG Tengfei, et al. Dual-domain aggregation Transformer for hyperspectral image super-resolution[J]. Journal of Northwest University (Natural Science Edition), 2026, 56(3): 551-560. DOI： 10.16152/j.cnki.xdxbzr.2026-03-008.

摘要

高光谱图像超分辨率旨在提升低空间分辨率高光谱图像的质量，对遥感、精准农业等领域的应用至关重要。尽管现有深度学习方法已经取得显著进展，但在刻画高光谱数据复杂的空间-光谱长程依赖关系以及充分挖掘空间与光谱特征互补性方面仍存在不足。为此，提出一种双域空间-光谱聚合Transformer网络（dual-domain spatial-spectral aggregation Transformer，DSAT）用于单高光谱图像超分辨率重建。DSAT围绕空间域与光谱域的双域建模与交互聚合展开设计：一方面，通过双向门控空间-光谱聚合模块（bidirectional gated spatial-spectral aggregation，BGSA）分别构建跨域空间自注意力和跨域光谱自注意力，其中，前者在局部窗口自注意力中引入全局上下文信息，以增强远距离空间位置之间的依赖建模，后者则聚焦光谱维度的全局关系，以精准刻画不同光谱波段之间的内在相关性；另一方面，在BGSA中嵌入自适应交互模块，从空间交互和通道交互2个层面对空谱特征进行自适应加权与融合，充分挖掘空间与光谱信息的互补性。3个基准数据集上的实验结果表明，该文所提方法在主客观质量方面均优于其他主流对比方法。

Abstract

Hyperspectral image super-resolution aims to enhance the spatial resolution of low-resolution hyper-spectral images and is critical for applications such as remote sensing and precision agriculture. Existing deep learning methods have made significant progress

but they still struggle to capture complex long-range spatial-spectral dependencies and to fully exploit complementary information between spatial and spectral features. To address these challenges

we propose a dual-domain spatial-spectral aggregation Transformer (DSAT) for single-image hyperspectral super-resolution. Central to DSAT is dual-domain modeling and interactive aggregation in the spatial and spectral domains. Concretely

a bidirectional gated spatial-spectral aggregation (BGSA) module constructs cross-domain spatial self-attention and cross-domain spectral self-attention. The spatial branch injects global context into local-window self-attention to strengthen long-range spatial dependency modeling

while the spectral branch focuses on global relationships along the spectral dimension to accurately capture intrinsic correlations among bands. Moreover

an adaptive interaction module is embedded within BGSA to perform adaptive weighting and fusion of spatial-spectral features via spatial interaction and channel interaction

thereby exploiting complementarity between spatial and spectral information. Experiments on three benchmark datasets show that the proposed method outperforms other competitors in both objective metrics and visual quality.

关键词

Keywords

references

XU Y C , ZHANG L F , DU B , et al . Hyperspectral anomaly detection based on machine learning: An overview [J ] . IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing , 2022 , 15 : 3351 - 3364 .

ZHANG Z , HUANG L H , WANG Q W , et al . UAV hyperspectral remote sensing image classification: A systematic review [J ] . IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing , 2025 , 18 : 3099 - 3124 .

XU X , LI J , HUANG X , et al . Multiple morphological component analysis based decomposition for remote sensing image classification [J ] . IEEE Transactions on Geoscience and Remote Sensing , 2016 , 54 ( 5 ): 3083 - 3102 .

LU G L , FEI B W . Medical hyperspectral imaging: A review [J ] . Journal of Biomedical Optics , 2014 , 19 ( 1 ): 010901 .

DIAN R W , FANG L Y , LI S T . Hyperspectral image super-resolution via non-local sparse tensor factorization [C ] // 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) . Honolulu, HI : IEEE , 2017 : 3862 - 3871 .

XU Y , WU Z B , CHANUSSOT J , et al . Nonlocal patch tensor sparse representation for hyperspectral image super-resolution [J ] . IEEE Transactions on Image Processing , 2019 , 28 ( 6 ): 3034 - 3047 .

YANG J C , WRIGHT J , HUANG T S , et al . Image super-resolution via sparse representation [J ] . IEEE Transactions on Image Processing , 2010 , 19 ( 11 ): 2861 - 2873 .

MEI S H , YUAN X , JI J Y , et al . Hyperspectral image super-resolution via convolutional neural network [C ] // 2017 IEEE International Conference on Image Processing (ICIP) . Beijing : IEEE , 2017 : 4297 - 4301 .

LI Q , WANG Q , LI X L . Exploring the relationship between 2D/3D convolution for hyperspectral image super-resolution [J ] . IEEE Transactions on Geoscience and Remote Sensing , 2021 , 59 ( 10 ): 8693 - 8703 .

WANG X Y , HU Q , JIANG J J , et al . A group-based embedding learning and integration network for hyperspectral image super-resolution [J ] . IEEE Transactions on Geoscience and Remote Sensing , 2022 , 60 : 5541416 .

LI Y , ZHANG L , DINGL C , et al . Single hyperspectral image super-resolution with grouped deep recursive residual network [C ] // 2018 IEEE Fourth International Conference on Multimedia Big Data (BigMM) . Xi’an : IEEE , 2018 : 1 - 4 .

JIANG J J , SUN H , LIU X M , et al . Learning spatial-spectral prior for super-resolution of hyperspectral imagery [J ] . IEEE Transactions on Computational Imaging , 2020 , 6 : 1082 - 1096 .

WANG X Y , MA J Y , JIANG J J . Hyperspectral image super-resolution via recurrent feedback embedding and spatial-spectral consistency regularization [J ] . IEEE Transactions on Geoscience and Remote Sensing , 2022 , 60 : 1 - 13 .

WANG H , WANG C , YUAN Y . Neighbor spectra maintenance and context affinity enhancement for single hyperspectral image super-resolution [J ] . IEEE Transactions on Geoscience and Remote Sensing , 2024 , 62 : 5515315 .

LIU Y T , HU J W , KANG X D , et al . Interactformer: Interactive transformer and CNN for hyperspectral image super-resolution [J ] . IEEE Transactions on Geoscience and Remote Sensing , 2022 , 60 : 5531715 .

ZHANG M J , ZHANG C , ZHANG Q M , et al . ESSAformer: Efficient Transformer for hyperspectral image super-resolution [C ] // 2023 IEEE/CVF International Conference on Computer Vision (ICCV) . Paris : IEEE , 2023 : 23016 - 23027 .

CHEN S , ZHANG L F , ZHANG L P . MSDformer: Multiscale deformable transformer for hyperspectral image super-resolution [J ] . IEEE Transactions on Geoscience and Remote Sensing , 2023 , 61 : 5525614 .

CHEN S , ZHANG L F , ZHANG L P . Cross-scope spatial-spectral information aggregation for hyperspectral image super-resolution [J ] . IEEE Transactions on Image Processing , 2024 , 33 : 5878 - 5891 .

HU Q , WANG X Y , JIANG J J , et al . Exploring the spectral prior for hyperspectral image super-resolution [J ] . IEEE Transactions on Image Processing , 2024 , 33 : 5260 - 5272 .

CHEN Z , ZHANG Y L , GU J J , et al . Dual aggregation transformer for image super-resolution [C ] // 2023 IEEE/CVF International Conference on Computer Vision (ICCV) . Paris : IEEE , 2023 : 12278 - 12287 .

GAMBA P . A collection of data for urban area characterization [C ] // 2004 IEEE International Geoscience and Remote Sensing Symposium . Anchorage : IEEE , 2004 : 69 - 72 .

YASUMA F , MITSUNAGA T , ISO D , et al . Generalized assorted pixel camera: Postcapture control of resolution, dynamic range, and spectrum [J ] . IEEE Transactions on Image Processing , 2010 , 19 ( 9 ): 2241 - 2253 .

岑奕，张立福，张霞，等 . 雄安新区马蹄湾村航空高光谱遥感影像分类数据集 [J ] . 遥感学报， 2020 , 24 ( 11 ): 1299 - 1306 .

CEN Y , ZHANG L F , ZHANG X , et al . Aerial hyperspectral remote sensing classification dataset of Xiongan New Area (MatiwanVillage) [J ] . Journal of Remote Sensing (Chinese) , 2020 , 24 ( 11 ): 1299 - 1306 .

SHI W Z , CABALLERO J , HUSZÁR F , et al . Realtime single image and video super-resolution using an efficient sub-pixel convolutional neural network [C ] // 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) . Las Vegas : IEEE , 2016 : 1874 - 1883 .

HOU J H , ZHU Z Y , HCU J H , et al . Deep posterior distribution-based embedding for hyperspectral image super-resolution [J ] . IEEE Transactions on Image Processing , 2022 , 31 : 5720 - 5732 .

XU Q , LIU S J , WANG J H , et al . AS 3 ITransUNet: Spatial-spectral interactive transformer U-Net with alternating sampling for hyperspectral image super-resolution [J ] . IEEE Transactions on Geoscience and Remote Sensing , 2023 , 61 : 1 - 13 .

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Joint-optimization deep network for biomedical images registration and segmentation based on the double-sampling data augmentation

Calligraphy image harmonization algorithm based on local light and shadow perception

Ancient clothing line drawing extraction based on Transformer two-stage strategy

Large-scale dataset generation technology for oracle object detection

Facial expression recognition in Swin Transformer by embedding hybrid attention mechanism

Related Author

QU Lei

LI Yuanyuan

LAI Chuxiao

YANG Bo

FENG Haoran

ZHANG Yong

WU Jun

XIAO Yun

Related Institution

Institute of Artificial Intelligence, Hefei Comprehensive National Science Center

The 38th Research Institute of China Electronics Technology Group Corporation

National Key Laboratory of Opto-Electronic Information Acquisition and Protection Technology, Anhui University

School of Information Science and Technology, Northwest University

National and Local Joint Engineering Research Center for Cultural Heritage Digitization, Northwest University

⁰