Fully Convolutional Spectral–Spatial Fusion Network Integrating Supervised Contrastive Learning for Hyperspectral Image Classification
Hyperspectral image classification using deep learning techniques has received great attention in recent years, considering the powerful spatial feature mining ability of deep learning. Fully convolutional network is an effective deep learning architecture that exploits spatial contextual informatio...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2023-01-01
|
| Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10265029/ |
| _version_ | 1797658058562732032 |
|---|---|
| author | Yifan Shen Ling Shi Ji Zhao Yuting Dong Lizhe Wang |
| author_facet | Yifan Shen Ling Shi Ji Zhao Yuting Dong Lizhe Wang |
| author_sort | Yifan Shen |
| collection | DOAJ |
| description | Hyperspectral image classification using deep learning techniques has received great attention in recent years, considering the powerful spatial feature mining ability of deep learning. Fully convolutional network is an effective deep learning architecture that exploits spatial contextual information through a hierarchical convolutional structure. However, it often ignores the relationships between samples of the same category and different categories within the global context. Therefore, a fully convolutional spectral–spatial fusion network based on supervised contrastive learning (FCSCL) is proposed for hyperspectral image classification to enhance the separability between different categories and class aggregation among the same category. In the FCSCL framework, the spectral–spatial fusion classification network is developed to capture subtle spectral variations and spatial patterns by adaptively fusing the features extracted by the spectral branch and spatial branch. To improve intraclass compactness and interclass separability, the SCL module is integrated into the FCSCL framework. The positive and negative sample pairs are constructed by the designed hard example pairs sampling strategy. These constructed sample pairs are used to guide the network to learn more discriminative feature representations that pixels of the same category are closer to each other and pixels of different categories are pushed further apart in the feature space. The experiments using three public hyperspectral datasets verify the effectiveness of the FCSCL algorithm, and the FCSCL method achieves better classification performance. |
| first_indexed | 2024-03-11T17:53:40Z |
| format | Article |
| id | doaj.art-f2a85b54487e465789fce9449016c7f4 |
| institution | Directory Open Access Journal |
| issn | 2151-1535 |
| language | English |
| last_indexed | 2024-03-11T17:53:40Z |
| publishDate | 2023-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| spelling | doaj.art-f2a85b54487e465789fce9449016c7f42023-10-17T23:00:14ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing2151-15352023-01-01169077908810.1109/JSTARS.2023.331958710265029Fully Convolutional Spectral–Spatial Fusion Network Integrating Supervised Contrastive Learning for Hyperspectral Image ClassificationYifan Shen0https://orcid.org/0009-0007-2779-521XLing Shi1https://orcid.org/0009-0003-5406-8967Ji Zhao2https://orcid.org/0000-0001-9039-2789Yuting Dong3https://orcid.org/0000-0002-8894-4318Lizhe Wang4https://orcid.org/0000-0003-2766-0845School of Computer Science, China University of Geosciences, Wuhan, ChinaSchool of Computer Science, China University of Geosciences, Wuhan, ChinaSchool of Computer Science, China University of Geosciences, Wuhan, ChinaSchool of Geography and Information Engineering, China University of Geosciences, Wuhan, ChinaSchool of Computer Science, China University of Geosciences, Wuhan, ChinaHyperspectral image classification using deep learning techniques has received great attention in recent years, considering the powerful spatial feature mining ability of deep learning. Fully convolutional network is an effective deep learning architecture that exploits spatial contextual information through a hierarchical convolutional structure. However, it often ignores the relationships between samples of the same category and different categories within the global context. Therefore, a fully convolutional spectral–spatial fusion network based on supervised contrastive learning (FCSCL) is proposed for hyperspectral image classification to enhance the separability between different categories and class aggregation among the same category. In the FCSCL framework, the spectral–spatial fusion classification network is developed to capture subtle spectral variations and spatial patterns by adaptively fusing the features extracted by the spectral branch and spatial branch. To improve intraclass compactness and interclass separability, the SCL module is integrated into the FCSCL framework. The positive and negative sample pairs are constructed by the designed hard example pairs sampling strategy. These constructed sample pairs are used to guide the network to learn more discriminative feature representations that pixels of the same category are closer to each other and pixels of different categories are pushed further apart in the feature space. The experiments using three public hyperspectral datasets verify the effectiveness of the FCSCL algorithm, and the FCSCL method achieves better classification performance.https://ieeexplore.ieee.org/document/10265029/Contrastive learningfully convolutional network (FCN)hyperspectral imageimage classification |
| spellingShingle | Yifan Shen Ling Shi Ji Zhao Yuting Dong Lizhe Wang Fully Convolutional Spectral–Spatial Fusion Network Integrating Supervised Contrastive Learning for Hyperspectral Image Classification IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing Contrastive learning fully convolutional network (FCN) hyperspectral image image classification |
| title | Fully Convolutional Spectral–Spatial Fusion Network Integrating Supervised Contrastive Learning for Hyperspectral Image Classification |
| title_full | Fully Convolutional Spectral–Spatial Fusion Network Integrating Supervised Contrastive Learning for Hyperspectral Image Classification |
| title_fullStr | Fully Convolutional Spectral–Spatial Fusion Network Integrating Supervised Contrastive Learning for Hyperspectral Image Classification |
| title_full_unstemmed | Fully Convolutional Spectral–Spatial Fusion Network Integrating Supervised Contrastive Learning for Hyperspectral Image Classification |
| title_short | Fully Convolutional Spectral–Spatial Fusion Network Integrating Supervised Contrastive Learning for Hyperspectral Image Classification |
| title_sort | fully convolutional spectral x2013 spatial fusion network integrating supervised contrastive learning for hyperspectral image classification |
| topic | Contrastive learning fully convolutional network (FCN) hyperspectral image image classification |
| url | https://ieeexplore.ieee.org/document/10265029/ |
| work_keys_str_mv | AT yifanshen fullyconvolutionalspectralx2013spatialfusionnetworkintegratingsupervisedcontrastivelearningforhyperspectralimageclassification AT lingshi fullyconvolutionalspectralx2013spatialfusionnetworkintegratingsupervisedcontrastivelearningforhyperspectralimageclassification AT jizhao fullyconvolutionalspectralx2013spatialfusionnetworkintegratingsupervisedcontrastivelearningforhyperspectralimageclassification AT yutingdong fullyconvolutionalspectralx2013spatialfusionnetworkintegratingsupervisedcontrastivelearningforhyperspectralimageclassification AT lizhewang fullyconvolutionalspectralx2013spatialfusionnetworkintegratingsupervisedcontrastivelearningforhyperspectralimageclassification |