An Unsupervised Band Selection Method via Contrastive Learning for Hyperspectral Images
Band selection (BS) is an efficacious approach to reduce hyperspectral information redundancy while preserving the physical meaning of hyperspectral images (HSIs). Recently, deep learning-based BS methods have received widespread interest due to their ability to model the nonlinear relationship betw...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-11-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/15/23/5495 |
| _version_ | 1797399648969687040 |
|---|---|
| author | Xiaorun Li Yufei Liu Ziqiang Hua Shuhan Chen |
| author_facet | Xiaorun Li Yufei Liu Ziqiang Hua Shuhan Chen |
| author_sort | Xiaorun Li |
| collection | DOAJ |
| description | Band selection (BS) is an efficacious approach to reduce hyperspectral information redundancy while preserving the physical meaning of hyperspectral images (HSIs). Recently, deep learning-based BS methods have received widespread interest due to their ability to model the nonlinear relationship between bands, with existing methods typically relying on generative algorithms. However, the process of generating images with pixel-level detail required by generative algorithm-based BS methods is computationally expensive. To alleviate this issue, we propose a contrastive learning-based unsupervised BS architecture, termed ContrastBS, in this article. With the help of contrastive learning, the proposed architecture avoids the costly generation step in pixel space by learning to distinguish data at the abstract semantic level of the feature space. Specifically, ContrastBS combines an attention mechanism with contrastive learning to extract the importance of each band. Furthermore, we design a novel loss function, which is able to constrain the symmetric loss while ensuring attention to the most valuable bands, for the contrastive learning-based BS network. Experimental results indicate that ContrastBS has excellent classification performance and competitive time cost compared to the comparison methods. |
| first_indexed | 2024-03-09T01:44:10Z |
| format | Article |
| id | doaj.art-90214972245149cb83b6bf04327ae7b1 |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-09T01:44:10Z |
| publishDate | 2023-11-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-90214972245149cb83b6bf04327ae7b12023-12-08T15:24:48ZengMDPI AGRemote Sensing2072-42922023-11-011523549510.3390/rs15235495An Unsupervised Band Selection Method via Contrastive Learning for Hyperspectral ImagesXiaorun Li0Yufei Liu1Ziqiang Hua2Shuhan Chen3College of Electrical Engineering, Zhejiang University, Hangzhou 310027, ChinaCollege of Electrical Engineering, Zhejiang University, Hangzhou 310027, ChinaCollege of Electrical Engineering, Zhejiang University, Hangzhou 310027, ChinaCollege of Electrical Engineering, Zhejiang University, Hangzhou 310027, ChinaBand selection (BS) is an efficacious approach to reduce hyperspectral information redundancy while preserving the physical meaning of hyperspectral images (HSIs). Recently, deep learning-based BS methods have received widespread interest due to their ability to model the nonlinear relationship between bands, with existing methods typically relying on generative algorithms. However, the process of generating images with pixel-level detail required by generative algorithm-based BS methods is computationally expensive. To alleviate this issue, we propose a contrastive learning-based unsupervised BS architecture, termed ContrastBS, in this article. With the help of contrastive learning, the proposed architecture avoids the costly generation step in pixel space by learning to distinguish data at the abstract semantic level of the feature space. Specifically, ContrastBS combines an attention mechanism with contrastive learning to extract the importance of each band. Furthermore, we design a novel loss function, which is able to constrain the symmetric loss while ensuring attention to the most valuable bands, for the contrastive learning-based BS network. Experimental results indicate that ContrastBS has excellent classification performance and competitive time cost compared to the comparison methods.https://www.mdpi.com/2072-4292/15/23/5495unsupervised band selectioncontrastive learninghyperspectral imageattention mechanism |
| spellingShingle | Xiaorun Li Yufei Liu Ziqiang Hua Shuhan Chen An Unsupervised Band Selection Method via Contrastive Learning for Hyperspectral Images Remote Sensing unsupervised band selection contrastive learning hyperspectral image attention mechanism |
| title | An Unsupervised Band Selection Method via Contrastive Learning for Hyperspectral Images |
| title_full | An Unsupervised Band Selection Method via Contrastive Learning for Hyperspectral Images |
| title_fullStr | An Unsupervised Band Selection Method via Contrastive Learning for Hyperspectral Images |
| title_full_unstemmed | An Unsupervised Band Selection Method via Contrastive Learning for Hyperspectral Images |
| title_short | An Unsupervised Band Selection Method via Contrastive Learning for Hyperspectral Images |
| title_sort | unsupervised band selection method via contrastive learning for hyperspectral images |
| topic | unsupervised band selection contrastive learning hyperspectral image attention mechanism |
| url | https://www.mdpi.com/2072-4292/15/23/5495 |
| work_keys_str_mv | AT xiaorunli anunsupervisedbandselectionmethodviacontrastivelearningforhyperspectralimages AT yufeiliu anunsupervisedbandselectionmethodviacontrastivelearningforhyperspectralimages AT ziqianghua anunsupervisedbandselectionmethodviacontrastivelearningforhyperspectralimages AT shuhanchen anunsupervisedbandselectionmethodviacontrastivelearningforhyperspectralimages AT xiaorunli unsupervisedbandselectionmethodviacontrastivelearningforhyperspectralimages AT yufeiliu unsupervisedbandselectionmethodviacontrastivelearningforhyperspectralimages AT ziqianghua unsupervisedbandselectionmethodviacontrastivelearningforhyperspectralimages AT shuhanchen unsupervisedbandselectionmethodviacontrastivelearningforhyperspectralimages |