A Novel 2D-3D CNN with Spectral-Spatial Multi-Scale Feature Fusion for Hyperspectral Image Classification

Multifarious hyperspectral image (HSI) classification methods based on convolutional neural networks (CNN) have been gradually proposed and achieve a promising classification performance. However, hyperspectral image classification still suffers from various challenges, including abundant redundant...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Dongxu Liu, Guangliang Han, Peixun Liu, Hang Yang, Xinglong Sun, Qingqing Li, Jiajia Wu
Formato:	article
Lenguaje:	EN
Publicado:	MDPI AG 2021
Materias:	hyperspectral image classification 2D-3D CNN multi-scale features multi-level features attention module Science Q
Acceso en línea:	https://doaj.org/article/e34cd057da4b4e6ebba8d152865c16d3
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:e34cd057da4b4e6ebba8d152865c16d3
record_format	dspace
spelling	oai:doaj.org-article:e34cd057da4b4e6ebba8d152865c16d32021-11-25T18:54:50ZA Novel 2D-3D CNN with Spectral-Spatial Multi-Scale Feature Fusion for Hyperspectral Image Classification10.3390/rs132246212072-4292https://doaj.org/article/e34cd057da4b4e6ebba8d152865c16d32021-11-01T00:00:00Zhttps://www.mdpi.com/2072-4292/13/22/4621https://doaj.org/toc/2072-4292Multifarious hyperspectral image (HSI) classification methods based on convolutional neural networks (CNN) have been gradually proposed and achieve a promising classification performance. However, hyperspectral image classification still suffers from various challenges, including abundant redundant information, insufficient spectral-spatial representation, irregular class distribution, and so forth. To address these issues, we propose a novel 2D-3D CNN with spectral-spatial multi-scale feature fusion for hyperspectral image classification, which consists of two feature extraction streams, a feature fusion module as well as a classification scheme. First, we employ two diverse backbone modules for feature representation, that is, the spectral feature and the spatial feature extraction streams. The former utilizes a hierarchical feature extraction module to capture multi-scale spectral features, while the latter extracts multi-stage spatial features by introducing a multi-level fusion structure. With these network units, the category attribute information of HSI can be fully excavated. Then, to output more complete and robust information for classification, a multi-scale spectral-spatial-semantic feature fusion module is presented based on a Decomposition-Reconstruction structure. Last of all, we innovate a classification scheme to lift the classification accuracy. Experimental results on three public datasets demonstrate that the proposed method outperforms the state-of-the-art methods.Dongxu LiuGuangliang HanPeixun LiuHang YangXinglong SunQingqing LiJiajia WuMDPI AGarticlehyperspectral image classification2D-3D CNNmulti-scale featuresmulti-level featuresattention moduleScienceQENRemote Sensing, Vol 13, Iss 4621, p 4621 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	hyperspectral image classification 2D-3D CNN multi-scale features multi-level features attention module Science Q
spellingShingle	hyperspectral image classification 2D-3D CNN multi-scale features multi-level features attention module Science Q Dongxu Liu Guangliang Han Peixun Liu Hang Yang Xinglong Sun Qingqing Li Jiajia Wu A Novel 2D-3D CNN with Spectral-Spatial Multi-Scale Feature Fusion for Hyperspectral Image Classification
description	Multifarious hyperspectral image (HSI) classification methods based on convolutional neural networks (CNN) have been gradually proposed and achieve a promising classification performance. However, hyperspectral image classification still suffers from various challenges, including abundant redundant information, insufficient spectral-spatial representation, irregular class distribution, and so forth. To address these issues, we propose a novel 2D-3D CNN with spectral-spatial multi-scale feature fusion for hyperspectral image classification, which consists of two feature extraction streams, a feature fusion module as well as a classification scheme. First, we employ two diverse backbone modules for feature representation, that is, the spectral feature and the spatial feature extraction streams. The former utilizes a hierarchical feature extraction module to capture multi-scale spectral features, while the latter extracts multi-stage spatial features by introducing a multi-level fusion structure. With these network units, the category attribute information of HSI can be fully excavated. Then, to output more complete and robust information for classification, a multi-scale spectral-spatial-semantic feature fusion module is presented based on a Decomposition-Reconstruction structure. Last of all, we innovate a classification scheme to lift the classification accuracy. Experimental results on three public datasets demonstrate that the proposed method outperforms the state-of-the-art methods.
format	article
author	Dongxu Liu Guangliang Han Peixun Liu Hang Yang Xinglong Sun Qingqing Li Jiajia Wu
author_facet	Dongxu Liu Guangliang Han Peixun Liu Hang Yang Xinglong Sun Qingqing Li Jiajia Wu
author_sort	Dongxu Liu
title	A Novel 2D-3D CNN with Spectral-Spatial Multi-Scale Feature Fusion for Hyperspectral Image Classification
title_short	A Novel 2D-3D CNN with Spectral-Spatial Multi-Scale Feature Fusion for Hyperspectral Image Classification
title_full	A Novel 2D-3D CNN with Spectral-Spatial Multi-Scale Feature Fusion for Hyperspectral Image Classification
title_fullStr	A Novel 2D-3D CNN with Spectral-Spatial Multi-Scale Feature Fusion for Hyperspectral Image Classification
title_full_unstemmed	A Novel 2D-3D CNN with Spectral-Spatial Multi-Scale Feature Fusion for Hyperspectral Image Classification
title_sort	novel 2d-3d cnn with spectral-spatial multi-scale feature fusion for hyperspectral image classification
publisher	MDPI AG
publishDate	2021
url	https://doaj.org/article/e34cd057da4b4e6ebba8d152865c16d3
work_keys_str_mv	AT dongxuliu anovel2d3dcnnwithspectralspatialmultiscalefeaturefusionforhyperspectralimageclassification AT guanglianghan anovel2d3dcnnwithspectralspatialmultiscalefeaturefusionforhyperspectralimageclassification AT peixunliu anovel2d3dcnnwithspectralspatialmultiscalefeaturefusionforhyperspectralimageclassification AT hangyang anovel2d3dcnnwithspectralspatialmultiscalefeaturefusionforhyperspectralimageclassification AT xinglongsun anovel2d3dcnnwithspectralspatialmultiscalefeaturefusionforhyperspectralimageclassification AT qingqingli anovel2d3dcnnwithspectralspatialmultiscalefeaturefusionforhyperspectralimageclassification AT jiajiawu anovel2d3dcnnwithspectralspatialmultiscalefeaturefusionforhyperspectralimageclassification AT dongxuliu novel2d3dcnnwithspectralspatialmultiscalefeaturefusionforhyperspectralimageclassification AT guanglianghan novel2d3dcnnwithspectralspatialmultiscalefeaturefusionforhyperspectralimageclassification AT peixunliu novel2d3dcnnwithspectralspatialmultiscalefeaturefusionforhyperspectralimageclassification AT hangyang novel2d3dcnnwithspectralspatialmultiscalefeaturefusionforhyperspectralimageclassification AT xinglongsun novel2d3dcnnwithspectralspatialmultiscalefeaturefusionforhyperspectralimageclassification AT qingqingli novel2d3dcnnwithspectralspatialmultiscalefeaturefusionforhyperspectralimageclassification AT jiajiawu novel2d3dcnnwithspectralspatialmultiscalefeaturefusionforhyperspectralimageclassification
_version_	1718410559748046848

A Novel 2D-3D CNN with Spectral-Spatial Multi-Scale Feature Fusion for Hyperspectral Image Classification

Ejemplares similares