Local Auxiliary Spatial–Spectral Decoupling Transformer Network for Cross-Scene Hyperspectral Image Classification

The feature-level domain alignment based on deep learning techniques has greatly improved the performance of unsupervised domain adaptation (UDA) for hyperspectral image (HSI) classification. However, most of these methods leverage convolutional neural networks to capture local features, overlooking...

Full description

Saved in:
Bibliographic Details
Main Authors: Qiusheng Chen, Zhuoqun Fang, Zhaokui Li, Qian Du, Shizhuo Deng, Tong Jia, Dongyue Chen
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/11023211/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849706187751161856
author Qiusheng Chen
Zhuoqun Fang
Zhaokui Li
Qian Du
Shizhuo Deng
Tong Jia
Dongyue Chen
author_facet Qiusheng Chen
Zhuoqun Fang
Zhaokui Li
Qian Du
Shizhuo Deng
Tong Jia
Dongyue Chen
author_sort Qiusheng Chen
collection DOAJ
description The feature-level domain alignment based on deep learning techniques has greatly improved the performance of unsupervised domain adaptation (UDA) for hyperspectral image (HSI) classification. However, most of these methods leverage convolutional neural networks to capture local features, overlooking the comparable spatial global (SaG) and spectral global (SeG) information shared by both the source and target domains. To overcome this issue, we propose a local auxiliary spatial–spectral decoupling transformer network to ease the learning of global domain-invariant information. The SaG and SeG features of HSIs are extracted through a dual-branch design, preventing the feature coupling of different dimensions. In order to compress the model’s parameter search space, a local auxiliary global feature extraction strategy is devised. First, local prior constraints are introduced by extracting primitive features using a convolutional intra-token embedding. Next, the extraction of global spatial and spectral information from these primitive features is effectively achieved using the self-attention mechanism. Finally, a dynamic feature fusion mechanism is devised that enables the model to focus on features more conducive to transfer while suppressing irrelevant features. By using only standard adversarial domain alignment, LASDT achieves the state-of-the-art performance, demonstrating the model’s superior capability in UDA for HSI classification.
format Article
id doaj-art-e08376d55f4f46c9adcd53e19002198e
institution DOAJ
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-e08376d55f4f46c9adcd53e19002198e2025-08-20T03:16:15ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing1939-14042151-15352025-01-0118147841480310.1109/JSTARS.2025.357636211023211Local Auxiliary Spatial–Spectral Decoupling Transformer Network for Cross-Scene Hyperspectral Image ClassificationQiusheng Chen0https://orcid.org/0000-0003-3978-0414Zhuoqun Fang1https://orcid.org/0000-0003-1259-5470Zhaokui Li2https://orcid.org/0000-0002-5331-7664Qian Du3https://orcid.org/0000-0001-8354-7500Shizhuo Deng4https://orcid.org/0000-0002-6863-8516Tong Jia5https://orcid.org/0000-0003-1424-798XDongyue Chen6https://orcid.org/0000-0003-0673-6767College of Information Science and Engineering, Northeastern University, Shenyang, ChinaCollege of Artificial Intelligence, Shenyang Aerospace University, Shenyang, ChinaSchool of Computer Science, Shenyang Aerospace University, Shenyang, ChinaDepartment of Electrical and Computer Engineering, Mississippi State University, Starkville, MS, USACollege of Information Science and Engineering, Northeastern University, Shenyang, ChinaCollege of Information Science and Engineering, Northeastern University, Shenyang, ChinaCollege of Information Science and Engineering, Northeastern University, Shenyang, ChinaThe feature-level domain alignment based on deep learning techniques has greatly improved the performance of unsupervised domain adaptation (UDA) for hyperspectral image (HSI) classification. However, most of these methods leverage convolutional neural networks to capture local features, overlooking the comparable spatial global (SaG) and spectral global (SeG) information shared by both the source and target domains. To overcome this issue, we propose a local auxiliary spatial–spectral decoupling transformer network to ease the learning of global domain-invariant information. The SaG and SeG features of HSIs are extracted through a dual-branch design, preventing the feature coupling of different dimensions. In order to compress the model’s parameter search space, a local auxiliary global feature extraction strategy is devised. First, local prior constraints are introduced by extracting primitive features using a convolutional intra-token embedding. Next, the extraction of global spatial and spectral information from these primitive features is effectively achieved using the self-attention mechanism. Finally, a dynamic feature fusion mechanism is devised that enables the model to focus on features more conducive to transfer while suppressing irrelevant features. By using only standard adversarial domain alignment, LASDT achieves the state-of-the-art performance, demonstrating the model’s superior capability in UDA for HSI classification.https://ieeexplore.ieee.org/document/11023211/Hyperspectral image (HSI) classificationtransformerunsupervised domain adaptation (UDA)
spellingShingle Qiusheng Chen
Zhuoqun Fang
Zhaokui Li
Qian Du
Shizhuo Deng
Tong Jia
Dongyue Chen
Local Auxiliary Spatial–Spectral Decoupling Transformer Network for Cross-Scene Hyperspectral Image Classification
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Hyperspectral image (HSI) classification
transformer
unsupervised domain adaptation (UDA)
title Local Auxiliary Spatial–Spectral Decoupling Transformer Network for Cross-Scene Hyperspectral Image Classification
title_full Local Auxiliary Spatial–Spectral Decoupling Transformer Network for Cross-Scene Hyperspectral Image Classification
title_fullStr Local Auxiliary Spatial–Spectral Decoupling Transformer Network for Cross-Scene Hyperspectral Image Classification
title_full_unstemmed Local Auxiliary Spatial–Spectral Decoupling Transformer Network for Cross-Scene Hyperspectral Image Classification
title_short Local Auxiliary Spatial–Spectral Decoupling Transformer Network for Cross-Scene Hyperspectral Image Classification
title_sort local auxiliary spatial x2013 spectral decoupling transformer network for cross scene hyperspectral image classification
topic Hyperspectral image (HSI) classification
transformer
unsupervised domain adaptation (UDA)
url https://ieeexplore.ieee.org/document/11023211/
work_keys_str_mv AT qiushengchen localauxiliaryspatialx2013spectraldecouplingtransformernetworkforcrossscenehyperspectralimageclassification
AT zhuoqunfang localauxiliaryspatialx2013spectraldecouplingtransformernetworkforcrossscenehyperspectralimageclassification
AT zhaokuili localauxiliaryspatialx2013spectraldecouplingtransformernetworkforcrossscenehyperspectralimageclassification
AT qiandu localauxiliaryspatialx2013spectraldecouplingtransformernetworkforcrossscenehyperspectralimageclassification
AT shizhuodeng localauxiliaryspatialx2013spectraldecouplingtransformernetworkforcrossscenehyperspectralimageclassification
AT tongjia localauxiliaryspatialx2013spectraldecouplingtransformernetworkforcrossscenehyperspectralimageclassification
AT dongyuechen localauxiliaryspatialx2013spectraldecouplingtransformernetworkforcrossscenehyperspectralimageclassification