Multibranch 3D-Dense Attention Network for Hyperspectral Image Classification
The convolutional neural network (CNN) is widely used in the task of hyperspectral image (HSI) classification. However, for the HSI of three-dimensional characteristics, the 2D CNN-based methods will result in losing spatial-spectral information. To solve this problem, this paper proposes a multi-br...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2022-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9815856/ |
| _version_ | 1811224006434488320 |
|---|---|
| author | Junru Yin Changsheng Qi Wei Huang Qiqiang Chen Jiantao Qu |
| author_facet | Junru Yin Changsheng Qi Wei Huang Qiqiang Chen Jiantao Qu |
| author_sort | Junru Yin |
| collection | DOAJ |
| description | The convolutional neural network (CNN) is widely used in the task of hyperspectral image (HSI) classification. However, for the HSI of three-dimensional characteristics, the 2D CNN-based methods will result in losing spatial-spectral information. To solve this problem, this paper proposes a multi-branch 3D-densely connected network for HSI classification. This network is able to reuse features to fully exploit the shallow spatial-spectral information of HSI. Meanwhile, the convolutional kernels of different sizes are used to extract multi-scale spatial-spectral features. Subsequently, spatial attention mechanisms are used to emphasize spatial features and increase the diversity of features. By introducing global average pooling instead of a fully connected layer, the number of parameters in the whole network will be reduced. In order to verify the performance of the proposed method, the experiment results conducted in the Indian Pines, the University of Pavia, Salinas Valley and Houston 2013 datasets show that the proposed method is better than the state-of-the-art methods. |
| first_indexed | 2024-04-12T08:42:59Z |
| format | Article |
| id | doaj.art-a5c3fd18a5b546f6a23a1ff500507515 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-04-12T08:42:59Z |
| publishDate | 2022-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-a5c3fd18a5b546f6a23a1ff5005075152022-12-22T03:39:48ZengIEEEIEEE Access2169-35362022-01-0110718867189810.1109/ACCESS.2022.31888539815856Multibranch 3D-Dense Attention Network for Hyperspectral Image ClassificationJunru Yin0https://orcid.org/0000-0002-7101-1140Changsheng Qi1Wei Huang2Qiqiang Chen3Jiantao Qu4College of Computer and Communication Engineering, Zhengzhou University of Light Industry, Zhengzhou, ChinaCollege of Computer and Communication Engineering, Zhengzhou University of Light Industry, Zhengzhou, ChinaCollege of Computer and Communication Engineering, Zhengzhou University of Light Industry, Zhengzhou, ChinaCollege of Computer and Communication Engineering, Zhengzhou University of Light Industry, Zhengzhou, ChinaCollege of Computer and Communication Engineering, Zhengzhou University of Light Industry, Zhengzhou, ChinaThe convolutional neural network (CNN) is widely used in the task of hyperspectral image (HSI) classification. However, for the HSI of three-dimensional characteristics, the 2D CNN-based methods will result in losing spatial-spectral information. To solve this problem, this paper proposes a multi-branch 3D-densely connected network for HSI classification. This network is able to reuse features to fully exploit the shallow spatial-spectral information of HSI. Meanwhile, the convolutional kernels of different sizes are used to extract multi-scale spatial-spectral features. Subsequently, spatial attention mechanisms are used to emphasize spatial features and increase the diversity of features. By introducing global average pooling instead of a fully connected layer, the number of parameters in the whole network will be reduced. In order to verify the performance of the proposed method, the experiment results conducted in the Indian Pines, the University of Pavia, Salinas Valley and Houston 2013 datasets show that the proposed method is better than the state-of-the-art methods.https://ieeexplore.ieee.org/document/9815856/Hyperspectral image classificationspatial-spectral featurespatial attention mechanism3D CNN |
| spellingShingle | Junru Yin Changsheng Qi Wei Huang Qiqiang Chen Jiantao Qu Multibranch 3D-Dense Attention Network for Hyperspectral Image Classification IEEE Access Hyperspectral image classification spatial-spectral feature spatial attention mechanism 3D CNN |
| title | Multibranch 3D-Dense Attention Network for Hyperspectral Image Classification |
| title_full | Multibranch 3D-Dense Attention Network for Hyperspectral Image Classification |
| title_fullStr | Multibranch 3D-Dense Attention Network for Hyperspectral Image Classification |
| title_full_unstemmed | Multibranch 3D-Dense Attention Network for Hyperspectral Image Classification |
| title_short | Multibranch 3D-Dense Attention Network for Hyperspectral Image Classification |
| title_sort | multibranch 3d dense attention network for hyperspectral image classification |
| topic | Hyperspectral image classification spatial-spectral feature spatial attention mechanism 3D CNN |
| url | https://ieeexplore.ieee.org/document/9815856/ |
| work_keys_str_mv | AT junruyin multibranch3ddenseattentionnetworkforhyperspectralimageclassification AT changshengqi multibranch3ddenseattentionnetworkforhyperspectralimageclassification AT weihuang multibranch3ddenseattentionnetworkforhyperspectralimageclassification AT qiqiangchen multibranch3ddenseattentionnetworkforhyperspectralimageclassification AT jiantaoqu multibranch3ddenseattentionnetworkforhyperspectralimageclassification |