Self-Supervised Deep Multi-Level Representation Learning Fusion-Based Maximum Entropy Subspace Clustering for Hyperspectral Band Selection
As one of the most important techniques for hyperspectral image dimensionality reduction, band selection has received considerable attention, whereas self-representation subspace clustering-based band selection algorithms have received quite a lot of attention with good effect. However, many of them...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-01-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/16/2/224 |
| _version_ | 1827371221690351616 |
|---|---|
| author | Yulei Wang Haipeng Ma Yuchao Yang Enyu Zhao Meiping Song Chunyan Yu |
| author_facet | Yulei Wang Haipeng Ma Yuchao Yang Enyu Zhao Meiping Song Chunyan Yu |
| author_sort | Yulei Wang |
| collection | DOAJ |
| description | As one of the most important techniques for hyperspectral image dimensionality reduction, band selection has received considerable attention, whereas self-representation subspace clustering-based band selection algorithms have received quite a lot of attention with good effect. However, many of them lack the self-supervision of representations and ignore the multi-level spectral–spatial information of HSI and the connectivity of subspaces. To this end, this paper proposes a novel self-supervised multi-level representation learning fusion-based maximum entropy subspace clustering (MLRLFMESC) method for hyperspectral band selection. Firstly, to learn multi-level spectral–spatial information, self-representation subspace clustering is embedded between the encoder layers of the deep-stacked convolutional autoencoder and its corresponding decoder layers, respectively, as multiple fully connected layers to achieve multi-level representation learning (MLRL). A new auxiliary task is constructed for multi-level representation learning and multi-level self-supervised training to improve its capability of representation. Then, a fusion model is designed to fuse the multi-level spectral–spatial information to obtain a more distinctive coefficient matrix for self-expression, where the maximum entropy regularization (MER) method is employed to promote connectivity and the uniform dense distribution of band elements in each subspace. Finally, subspace clustering is conducted to obtain the final band subset. Experiments have been conducted on three hyperspectral datasets, and the corresponding results show that the proposed MLRLFMESC algorithm significantly outperforms several other band selection methods in classification performance. |
| first_indexed | 2024-03-08T10:36:08Z |
| format | Article |
| id | doaj.art-263fafe0774444b899e925990d08f2a8 |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-08T10:36:08Z |
| publishDate | 2024-01-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-263fafe0774444b899e925990d08f2a82024-01-26T18:15:47ZengMDPI AGRemote Sensing2072-42922024-01-0116222410.3390/rs16020224Self-Supervised Deep Multi-Level Representation Learning Fusion-Based Maximum Entropy Subspace Clustering for Hyperspectral Band SelectionYulei Wang0Haipeng Ma1Yuchao Yang2Enyu Zhao3Meiping Song4Chunyan Yu5Information Science and Technology College, Dalian Maritime University, Dalian 116026, ChinaInformation Science and Technology College, Dalian Maritime University, Dalian 116026, ChinaInformation Science and Technology College, Dalian Maritime University, Dalian 116026, ChinaInformation Science and Technology College, Dalian Maritime University, Dalian 116026, ChinaInformation Science and Technology College, Dalian Maritime University, Dalian 116026, ChinaInformation Science and Technology College, Dalian Maritime University, Dalian 116026, ChinaAs one of the most important techniques for hyperspectral image dimensionality reduction, band selection has received considerable attention, whereas self-representation subspace clustering-based band selection algorithms have received quite a lot of attention with good effect. However, many of them lack the self-supervision of representations and ignore the multi-level spectral–spatial information of HSI and the connectivity of subspaces. To this end, this paper proposes a novel self-supervised multi-level representation learning fusion-based maximum entropy subspace clustering (MLRLFMESC) method for hyperspectral band selection. Firstly, to learn multi-level spectral–spatial information, self-representation subspace clustering is embedded between the encoder layers of the deep-stacked convolutional autoencoder and its corresponding decoder layers, respectively, as multiple fully connected layers to achieve multi-level representation learning (MLRL). A new auxiliary task is constructed for multi-level representation learning and multi-level self-supervised training to improve its capability of representation. Then, a fusion model is designed to fuse the multi-level spectral–spatial information to obtain a more distinctive coefficient matrix for self-expression, where the maximum entropy regularization (MER) method is employed to promote connectivity and the uniform dense distribution of band elements in each subspace. Finally, subspace clustering is conducted to obtain the final band subset. Experiments have been conducted on three hyperspectral datasets, and the corresponding results show that the proposed MLRLFMESC algorithm significantly outperforms several other band selection methods in classification performance.https://www.mdpi.com/2072-4292/16/2/224hyperspectral imageryband selectionmulti-level representation learningmulti-level self-supervised learningmaximum entropy subspace clustering |
| spellingShingle | Yulei Wang Haipeng Ma Yuchao Yang Enyu Zhao Meiping Song Chunyan Yu Self-Supervised Deep Multi-Level Representation Learning Fusion-Based Maximum Entropy Subspace Clustering for Hyperspectral Band Selection Remote Sensing hyperspectral imagery band selection multi-level representation learning multi-level self-supervised learning maximum entropy subspace clustering |
| title | Self-Supervised Deep Multi-Level Representation Learning Fusion-Based Maximum Entropy Subspace Clustering for Hyperspectral Band Selection |
| title_full | Self-Supervised Deep Multi-Level Representation Learning Fusion-Based Maximum Entropy Subspace Clustering for Hyperspectral Band Selection |
| title_fullStr | Self-Supervised Deep Multi-Level Representation Learning Fusion-Based Maximum Entropy Subspace Clustering for Hyperspectral Band Selection |
| title_full_unstemmed | Self-Supervised Deep Multi-Level Representation Learning Fusion-Based Maximum Entropy Subspace Clustering for Hyperspectral Band Selection |
| title_short | Self-Supervised Deep Multi-Level Representation Learning Fusion-Based Maximum Entropy Subspace Clustering for Hyperspectral Band Selection |
| title_sort | self supervised deep multi level representation learning fusion based maximum entropy subspace clustering for hyperspectral band selection |
| topic | hyperspectral imagery band selection multi-level representation learning multi-level self-supervised learning maximum entropy subspace clustering |
| url | https://www.mdpi.com/2072-4292/16/2/224 |
| work_keys_str_mv | AT yuleiwang selfsuperviseddeepmultilevelrepresentationlearningfusionbasedmaximumentropysubspaceclusteringforhyperspectralbandselection AT haipengma selfsuperviseddeepmultilevelrepresentationlearningfusionbasedmaximumentropysubspaceclusteringforhyperspectralbandselection AT yuchaoyang selfsuperviseddeepmultilevelrepresentationlearningfusionbasedmaximumentropysubspaceclusteringforhyperspectralbandselection AT enyuzhao selfsuperviseddeepmultilevelrepresentationlearningfusionbasedmaximumentropysubspaceclusteringforhyperspectralbandselection AT meipingsong selfsuperviseddeepmultilevelrepresentationlearningfusionbasedmaximumentropysubspaceclusteringforhyperspectralbandselection AT chunyanyu selfsuperviseddeepmultilevelrepresentationlearningfusionbasedmaximumentropysubspaceclusteringforhyperspectralbandselection |