Lightweight Spatial–Spectral Shift Module With Multihead MambaOut for Hyperspectral Image Classification

In hyperspectral images, the high dimensionality of spectral data often leads to redundant spectral information, making it difficult to extract features. Two-dimensional CNNs fail to effec-tively extract spatial and spectral information, and deploying three-dimensional CNNs on microprocessors is cha...

Full description

Saved in:
Bibliographic Details
Main Authors: Yi Liu, Yanjun Zhang, Yu Guo, Yunchao Li
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/10767195/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850029813161525248
author Yi Liu
Yanjun Zhang
Yu Guo
Yunchao Li
author_facet Yi Liu
Yanjun Zhang
Yu Guo
Yunchao Li
author_sort Yi Liu
collection DOAJ
description In hyperspectral images, the high dimensionality of spectral data often leads to redundant spectral information, making it difficult to extract features. Two-dimensional CNNs fail to effec-tively extract spatial and spectral information, and deploying three-dimensional CNNs on microprocessors is challenging as these net-works consume excessive resources. If graph convolutional networks (GCN) are adopted, most networks employ superpixel segmentation for HSI classification. However, this approach tends to overlook pixel-level features and thus fails to achieve fine classification.To efficiently extract spectral features while reducing resource consumption, we proposed the spectral shift module (SPCSM). This module extracts features by circularly shifting spectral information. It emphasizes the internal correlations within the spectral and offers the advantage of fewer parameters. Based on the SPCSM, we designed the spatial&#x2013;spectral shift module (S<sup>2</sup>SM). It models and analyzes correlations between spatial and spectral data, facilitating the extraction of spatial and spectral features. Additionally, it addresses the issues of redundancy in shallow information. To enable deployment on microprocessors and effectively classify hyperspectral images, we optimized the MambaOut network by integrating the S<sup>2</sup>SM with a multihead MambaOut (LS<sup>2</sup>SM-MHMambaOut). Experiments demonstrated that our network struck a favorable balance between classification accuracy and model complexity. Therefore, it is promising to be integrated into microprocessors.
format Article
id doaj-art-e3d278e8a8854c84bb51dc554d046655
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-e3d278e8a8854c84bb51dc554d0466552025-08-20T02:59:24ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing1939-14042151-15352025-01-011892193410.1109/JSTARS.2024.350598410767195Lightweight Spatial&#x2013;Spectral Shift Module With Multihead MambaOut for Hyperspectral Image ClassificationYi Liu0https://orcid.org/0009-0003-5463-4711Yanjun Zhang1https://orcid.org/0009-0006-9947-8881Yu Guo2Yunchao Li3State Key Laboratory of Dynamic Measurement Techno-logy, North University of China, Yaiyuan, ChinaState Key Laboratory of Dynamic Measurement Technology, North University of China, Yaiyuan, ChinaBeijing Electro-mechanical Engineering Institute, Beijing, ChinaNorinco Group Test and Measuring Academy, Xi&#x0027;an, ChinaIn hyperspectral images, the high dimensionality of spectral data often leads to redundant spectral information, making it difficult to extract features. Two-dimensional CNNs fail to effec-tively extract spatial and spectral information, and deploying three-dimensional CNNs on microprocessors is challenging as these net-works consume excessive resources. If graph convolutional networks (GCN) are adopted, most networks employ superpixel segmentation for HSI classification. However, this approach tends to overlook pixel-level features and thus fails to achieve fine classification.To efficiently extract spectral features while reducing resource consumption, we proposed the spectral shift module (SPCSM). This module extracts features by circularly shifting spectral information. It emphasizes the internal correlations within the spectral and offers the advantage of fewer parameters. Based on the SPCSM, we designed the spatial&#x2013;spectral shift module (S<sup>2</sup>SM). It models and analyzes correlations between spatial and spectral data, facilitating the extraction of spatial and spectral features. Additionally, it addresses the issues of redundancy in shallow information. To enable deployment on microprocessors and effectively classify hyperspectral images, we optimized the MambaOut network by integrating the S<sup>2</sup>SM with a multihead MambaOut (LS<sup>2</sup>SM-MHMambaOut). Experiments demonstrated that our network struck a favorable balance between classification accuracy and model complexity. Therefore, it is promising to be integrated into microprocessors.https://ieeexplore.ieee.org/document/10767195/Hyperspectral images (HSI)MambaOutspatial–spectral shift module
spellingShingle Yi Liu
Yanjun Zhang
Yu Guo
Yunchao Li
Lightweight Spatial&#x2013;Spectral Shift Module With Multihead MambaOut for Hyperspectral Image Classification
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Hyperspectral images (HSI)
MambaOut
spatial–spectral shift module
title Lightweight Spatial&#x2013;Spectral Shift Module With Multihead MambaOut for Hyperspectral Image Classification
title_full Lightweight Spatial&#x2013;Spectral Shift Module With Multihead MambaOut for Hyperspectral Image Classification
title_fullStr Lightweight Spatial&#x2013;Spectral Shift Module With Multihead MambaOut for Hyperspectral Image Classification
title_full_unstemmed Lightweight Spatial&#x2013;Spectral Shift Module With Multihead MambaOut for Hyperspectral Image Classification
title_short Lightweight Spatial&#x2013;Spectral Shift Module With Multihead MambaOut for Hyperspectral Image Classification
title_sort lightweight spatial x2013 spectral shift module with multihead mambaout for hyperspectral image classification
topic Hyperspectral images (HSI)
MambaOut
spatial–spectral shift module
url https://ieeexplore.ieee.org/document/10767195/
work_keys_str_mv AT yiliu lightweightspatialx2013spectralshiftmodulewithmultiheadmambaoutforhyperspectralimageclassification
AT yanjunzhang lightweightspatialx2013spectralshiftmodulewithmultiheadmambaoutforhyperspectralimageclassification
AT yuguo lightweightspatialx2013spectralshiftmodulewithmultiheadmambaoutforhyperspectralimageclassification
AT yunchaoli lightweightspatialx2013spectralshiftmodulewithmultiheadmambaoutforhyperspectralimageclassification