Multiscale Spatial-Spectral CNN-Transformer Network for Hyperspectral Image Super-Resolution

Remarkable strides have been made in super-resolution methods based on deep learning for hyperspectral images (HSIs), which are capable of enhancing the spatial resolution. However, these methods predominantly focus on capturing local features using convolutional neural networks (CNNs), neglecting t...

Full description

Saved in:
Bibliographic Details
Main Authors: Jiayang Zhang, Hongjia Qu, Junhao Jia, Yaowei Li, Bo Jiang, Xiaoxuan Chen, Jinye Peng
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10980410/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850128648035631104
author Jiayang Zhang
Hongjia Qu
Junhao Jia
Yaowei Li
Bo Jiang
Xiaoxuan Chen
Jinye Peng
author_facet Jiayang Zhang
Hongjia Qu
Junhao Jia
Yaowei Li
Bo Jiang
Xiaoxuan Chen
Jinye Peng
author_sort Jiayang Zhang
collection DOAJ
description Remarkable strides have been made in super-resolution methods based on deep learning for hyperspectral images (HSIs), which are capable of enhancing the spatial resolution. However, these methods predominantly focus on capturing local features using convolutional neural networks (CNNs), neglecting the comprehensive utilization of global spatial-spectral information. To address this limitation, we innovatively propose a multiscale spatial-spectral CNN-transformer network for hyperspectral image super resolution, namely, MSHSR. MSHSR not only applies the local spatial-spectral characteristics but also innovatively facilitates the collaborative exploration and application of spatial details and spectral data globally. Specifically, we first design a multiscale spatial-spectral fusion module, which integrates dilated-convolution parallel branches and a hybrid spectral attention mechanism to address the strong local correlations in HSIs, effectively capturing and fusing multiscale local spatial-spectral information. Furthermore, in order to fully exploit the global contextual consistency in HSIs, we introduce a sparse spectral transformer module. This module processes the previously obtained local spatial-spectral features, thoroughly exploring the elaborate global interrelationship and long-range dependencies among different spectral bands through a coarse-to-fine strategy. Extensive experimental results on three hyperspectral datasets demonstrate the superior performance of our method, outperforming comparison methods both in quantitative metrics and visual performance.
format Article
id doaj-art-e2095064638742e8b4176041699a7935
institution OA Journals
issn 1939-1404
2151-1535
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
spelling doaj-art-e2095064638742e8b4176041699a79352025-08-20T02:33:13ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing1939-14042151-15352025-01-0118121161213210.1109/JSTARS.2025.356584010980410Multiscale Spatial-Spectral CNN-Transformer Network for Hyperspectral Image Super-ResolutionJiayang Zhang0Hongjia Qu1Junhao Jia2Yaowei Li3https://orcid.org/0000-0002-9271-2974Bo Jiang4https://orcid.org/0000-0002-4066-1802Xiaoxuan Chen5https://orcid.org/0000-0002-2029-2448Jinye Peng6https://orcid.org/0000-0002-6725-1387School of Information Science and Technology, Northwest University, Xi'an, ChinaSchool of Information Science and Technology, Northwest University, Xi'an, ChinaSchool of Information Science and Technology, Northwest University, Xi'an, ChinaSchool of Information Science and Technology, Northwest University, Xi'an, ChinaSchool of Information Science and Technology, Northwest University, Xi'an, ChinaSchool of Information Science and Technology, Northwest University, Xi'an, ChinaSchool of Information Science and Technology, Northwest University, Xi'an, ChinaRemarkable strides have been made in super-resolution methods based on deep learning for hyperspectral images (HSIs), which are capable of enhancing the spatial resolution. However, these methods predominantly focus on capturing local features using convolutional neural networks (CNNs), neglecting the comprehensive utilization of global spatial-spectral information. To address this limitation, we innovatively propose a multiscale spatial-spectral CNN-transformer network for hyperspectral image super resolution, namely, MSHSR. MSHSR not only applies the local spatial-spectral characteristics but also innovatively facilitates the collaborative exploration and application of spatial details and spectral data globally. Specifically, we first design a multiscale spatial-spectral fusion module, which integrates dilated-convolution parallel branches and a hybrid spectral attention mechanism to address the strong local correlations in HSIs, effectively capturing and fusing multiscale local spatial-spectral information. Furthermore, in order to fully exploit the global contextual consistency in HSIs, we introduce a sparse spectral transformer module. This module processes the previously obtained local spatial-spectral features, thoroughly exploring the elaborate global interrelationship and long-range dependencies among different spectral bands through a coarse-to-fine strategy. Extensive experimental results on three hyperspectral datasets demonstrate the superior performance of our method, outperforming comparison methods both in quantitative metrics and visual performance.https://ieeexplore.ieee.org/document/10980410/Hyperspectral image (HSI)self-attentionsuper-resolution (SR)transformer
spellingShingle Jiayang Zhang
Hongjia Qu
Junhao Jia
Yaowei Li
Bo Jiang
Xiaoxuan Chen
Jinye Peng
Multiscale Spatial-Spectral CNN-Transformer Network for Hyperspectral Image Super-Resolution
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Hyperspectral image (HSI)
self-attention
super-resolution (SR)
transformer
title Multiscale Spatial-Spectral CNN-Transformer Network for Hyperspectral Image Super-Resolution
title_full Multiscale Spatial-Spectral CNN-Transformer Network for Hyperspectral Image Super-Resolution
title_fullStr Multiscale Spatial-Spectral CNN-Transformer Network for Hyperspectral Image Super-Resolution
title_full_unstemmed Multiscale Spatial-Spectral CNN-Transformer Network for Hyperspectral Image Super-Resolution
title_short Multiscale Spatial-Spectral CNN-Transformer Network for Hyperspectral Image Super-Resolution
title_sort multiscale spatial spectral cnn transformer network for hyperspectral image super resolution
topic Hyperspectral image (HSI)
self-attention
super-resolution (SR)
transformer
url https://ieeexplore.ieee.org/document/10980410/
work_keys_str_mv AT jiayangzhang multiscalespatialspectralcnntransformernetworkforhyperspectralimagesuperresolution
AT hongjiaqu multiscalespatialspectralcnntransformernetworkforhyperspectralimagesuperresolution
AT junhaojia multiscalespatialspectralcnntransformernetworkforhyperspectralimagesuperresolution
AT yaoweili multiscalespatialspectralcnntransformernetworkforhyperspectralimagesuperresolution
AT bojiang multiscalespatialspectralcnntransformernetworkforhyperspectralimagesuperresolution
AT xiaoxuanchen multiscalespatialspectralcnntransformernetworkforhyperspectralimagesuperresolution
AT jinyepeng multiscalespatialspectralcnntransformernetworkforhyperspectralimagesuperresolution