A Pixel Cluster CNN and Spectral-Spatial Fusion Algorithm for Hyperspectral Image Classification With Small-Size Training Samples
Convolutional neural networks (CNNs) can automatically learn features from the hyperspectral image (HSI) data, avoiding the difficulty of manually extracting features. However, the number of training samples for the classification of HSIs is always limited, making it difficult for CNN to obtain effe...
Main Authors: | , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2021-01-01
|
Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/9388863/ |
id |
doaj-111603a4aa184095982f0e18ec5cd01c |
---|---|
record_format |
Article |
spelling |
doaj-111603a4aa184095982f0e18ec5cd01c2021-06-03T23:05:31ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing2151-15352021-01-01144101411410.1109/JSTARS.2021.30688649388863A Pixel Cluster CNN and Spectral-Spatial Fusion Algorithm for Hyperspectral Image Classification With Small-Size Training SamplesShuxian Dong0Yinghui Quan1https://orcid.org/0000-0001-6541-9441Wei Feng2https://orcid.org/0000-0003-1907-2664Gabriel Dauphin3https://orcid.org/0000-0002-0677-6702Lianru Gao4https://orcid.org/0000-0003-3888-8124Mengdao Xing5https://orcid.org/0000-0002-4084-0915Department of Remote Sensing Science, and Technology, School of Electronic Engineering, Xidian University, Xi’an, ChinaDepartment of Remote Sensing Science, and Technology, School of Electronic Engineering, Xidian University, Xi’an, ChinaDepartment of Remote Sensing Science, and Technology, School of Electronic Engineering, Xidian University, Xi’an, ChinaLaboratory of Information Processing, and Transmission, Institut Galilée, University Paris XIII, Villetaneuse, FranceKey Laboratory of Digital Earth Science, Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing, ChinaAcademy of Advanced Interdisciplinary Research, Xidian University, Xi’an, ChinaConvolutional neural networks (CNNs) can automatically learn features from the hyperspectral image (HSI) data, avoiding the difficulty of manually extracting features. However, the number of training samples for the classification of HSIs is always limited, making it difficult for CNN to obtain effective features and resulting in low classification accuracy. To solve this problem, a pixel cluster CNN and spectral-spatial fusion (SSF) algorithm for hyperspectral image classification with small-size training samples is proposed in this article. First, spatial information is extracted by the gray level co-occurrence matrix. Then, spatial information and spectral information are fused by means of bands superposition, forming spectral-spatial features. To expand the number of training samples, the pixels after SSF are combined into pixel clusters according to a certain rule. Finally, a CNN framework is utilized to extract effective features from the pixel clusters. Experiments based on three standard HSIs demonstrate that the proposed algorithm can get better performance than the conventional CNN and also outperforms other studied algorithms in the case of small training set.https://ieeexplore.ieee.org/document/9388863/Convolutional neural network (CNN)hyperspectral image classification (HSIC)pixel clustersmall training setspectral-spatial fusion (SSF) |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Shuxian Dong Yinghui Quan Wei Feng Gabriel Dauphin Lianru Gao Mengdao Xing |
spellingShingle |
Shuxian Dong Yinghui Quan Wei Feng Gabriel Dauphin Lianru Gao Mengdao Xing A Pixel Cluster CNN and Spectral-Spatial Fusion Algorithm for Hyperspectral Image Classification With Small-Size Training Samples IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing Convolutional neural network (CNN) hyperspectral image classification (HSIC) pixel cluster small training set spectral-spatial fusion (SSF) |
author_facet |
Shuxian Dong Yinghui Quan Wei Feng Gabriel Dauphin Lianru Gao Mengdao Xing |
author_sort |
Shuxian Dong |
title |
A Pixel Cluster CNN and Spectral-Spatial Fusion Algorithm for Hyperspectral Image Classification With Small-Size Training Samples |
title_short |
A Pixel Cluster CNN and Spectral-Spatial Fusion Algorithm for Hyperspectral Image Classification With Small-Size Training Samples |
title_full |
A Pixel Cluster CNN and Spectral-Spatial Fusion Algorithm for Hyperspectral Image Classification With Small-Size Training Samples |
title_fullStr |
A Pixel Cluster CNN and Spectral-Spatial Fusion Algorithm for Hyperspectral Image Classification With Small-Size Training Samples |
title_full_unstemmed |
A Pixel Cluster CNN and Spectral-Spatial Fusion Algorithm for Hyperspectral Image Classification With Small-Size Training Samples |
title_sort |
pixel cluster cnn and spectral-spatial fusion algorithm for hyperspectral image classification with small-size training samples |
publisher |
IEEE |
series |
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
issn |
2151-1535 |
publishDate |
2021-01-01 |
description |
Convolutional neural networks (CNNs) can automatically learn features from the hyperspectral image (HSI) data, avoiding the difficulty of manually extracting features. However, the number of training samples for the classification of HSIs is always limited, making it difficult for CNN to obtain effective features and resulting in low classification accuracy. To solve this problem, a pixel cluster CNN and spectral-spatial fusion (SSF) algorithm for hyperspectral image classification with small-size training samples is proposed in this article. First, spatial information is extracted by the gray level co-occurrence matrix. Then, spatial information and spectral information are fused by means of bands superposition, forming spectral-spatial features. To expand the number of training samples, the pixels after SSF are combined into pixel clusters according to a certain rule. Finally, a CNN framework is utilized to extract effective features from the pixel clusters. Experiments based on three standard HSIs demonstrate that the proposed algorithm can get better performance than the conventional CNN and also outperforms other studied algorithms in the case of small training set. |
topic |
Convolutional neural network (CNN) hyperspectral image classification (HSIC) pixel cluster small training set spectral-spatial fusion (SSF) |
url |
https://ieeexplore.ieee.org/document/9388863/ |
work_keys_str_mv |
AT shuxiandong apixelclustercnnandspectralspatialfusionalgorithmforhyperspectralimageclassificationwithsmallsizetrainingsamples AT yinghuiquan apixelclustercnnandspectralspatialfusionalgorithmforhyperspectralimageclassificationwithsmallsizetrainingsamples AT weifeng apixelclustercnnandspectralspatialfusionalgorithmforhyperspectralimageclassificationwithsmallsizetrainingsamples AT gabrieldauphin apixelclustercnnandspectralspatialfusionalgorithmforhyperspectralimageclassificationwithsmallsizetrainingsamples AT lianrugao apixelclustercnnandspectralspatialfusionalgorithmforhyperspectralimageclassificationwithsmallsizetrainingsamples AT mengdaoxing apixelclustercnnandspectralspatialfusionalgorithmforhyperspectralimageclassificationwithsmallsizetrainingsamples AT shuxiandong pixelclustercnnandspectralspatialfusionalgorithmforhyperspectralimageclassificationwithsmallsizetrainingsamples AT yinghuiquan pixelclustercnnandspectralspatialfusionalgorithmforhyperspectralimageclassificationwithsmallsizetrainingsamples AT weifeng pixelclustercnnandspectralspatialfusionalgorithmforhyperspectralimageclassificationwithsmallsizetrainingsamples AT gabrieldauphin pixelclustercnnandspectralspatialfusionalgorithmforhyperspectralimageclassificationwithsmallsizetrainingsamples AT lianrugao pixelclustercnnandspectralspatialfusionalgorithmforhyperspectralimageclassificationwithsmallsizetrainingsamples AT mengdaoxing pixelclustercnnandspectralspatialfusionalgorithmforhyperspectralimageclassificationwithsmallsizetrainingsamples |
_version_ |
1721398640996515840 |