Spectral-Spatial-Sensorial Attention Network with Controllable Factors for Hyperspectral Image Classification
Hyperspectral image (HSI) classification aims to recognize categories of objects based on spectral–spatial features and has been used in a wide range of real-world application areas. Attention mechanisms are widely used in HSI classification for their ability to focus on important information in ima...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-04-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/16/7/1253 |
| _version_ | 1797212055727505408 |
|---|---|
| author | Sheng Li Mingwei Wang Chong Cheng Xianjun Gao Zhiwei Ye Wei Liu |
| author_facet | Sheng Li Mingwei Wang Chong Cheng Xianjun Gao Zhiwei Ye Wei Liu |
| author_sort | Sheng Li |
| collection | DOAJ |
| description | Hyperspectral image (HSI) classification aims to recognize categories of objects based on spectral–spatial features and has been used in a wide range of real-world application areas. Attention mechanisms are widely used in HSI classification for their ability to focus on important information in images automatically. However, due to the approximate spectral–spatial features in HSI, mainstream attention mechanisms are difficult to accurately distinguish the small difference, which limits the classification accuracy. To overcome this problem, a spectral–spatial-sensorial attention network (S<sup>3</sup>AN) with controllable factors is proposed to efficiently recognize different objects. Specifically, two controllable factors, dynamic exponential pooling (DE-Pooling) and adaptive convolution (Adapt-Conv), are designed to enlarge the difference in approximate features and enhance the attention weight interaction. Then, attention mechanisms with controllable factors are utilized to build the redundancy reduction module (RRM), feature learning module (FLM), and label prediction module (LPM) to process HSI spectral–spatial features. The RRM utilizes the spectral attention mechanism to select representative band combinations, and the FLM introduces the spatial attention mechanism to highlight important objects. Furthermore, the sensorial attention mechanism extracts location and category information in a pseudo label to guide the LPM for label prediction and avoid details from being ignored. Experimental results on three public HSI datasets show that the proposed method is able to accurately recognize different objects with an overall accuracy (OA) of 98.69%, 98.89%, and 97.56%, respectively. |
| first_indexed | 2024-04-24T10:36:18Z |
| format | Article |
| id | doaj.art-fc8954118faf420393c6244a7e0d9a3f |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-04-24T10:36:18Z |
| publishDate | 2024-04-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-fc8954118faf420393c6244a7e0d9a3f2024-04-12T13:25:46ZengMDPI AGRemote Sensing2072-42922024-04-01167125310.3390/rs16071253Spectral-Spatial-Sensorial Attention Network with Controllable Factors for Hyperspectral Image ClassificationSheng Li0Mingwei Wang1Chong Cheng2Xianjun Gao3Zhiwei Ye4Wei Liu5School of Computer Science, Hubei University of Technology, Wuhan 430068, ChinaSchool of Computer Science, Hubei University of Technology, Wuhan 430068, ChinaSchool of Computer Science, Hubei University of Technology, Wuhan 430068, ChinaSchool of Geosciences, Yangtze University, Wuhan 430100, ChinaSchool of Computer Science, Hubei University of Technology, Wuhan 430068, ChinaInstitute of Geological Survey, China University of Geosciences, Wuhan 430074, ChinaHyperspectral image (HSI) classification aims to recognize categories of objects based on spectral–spatial features and has been used in a wide range of real-world application areas. Attention mechanisms are widely used in HSI classification for their ability to focus on important information in images automatically. However, due to the approximate spectral–spatial features in HSI, mainstream attention mechanisms are difficult to accurately distinguish the small difference, which limits the classification accuracy. To overcome this problem, a spectral–spatial-sensorial attention network (S<sup>3</sup>AN) with controllable factors is proposed to efficiently recognize different objects. Specifically, two controllable factors, dynamic exponential pooling (DE-Pooling) and adaptive convolution (Adapt-Conv), are designed to enlarge the difference in approximate features and enhance the attention weight interaction. Then, attention mechanisms with controllable factors are utilized to build the redundancy reduction module (RRM), feature learning module (FLM), and label prediction module (LPM) to process HSI spectral–spatial features. The RRM utilizes the spectral attention mechanism to select representative band combinations, and the FLM introduces the spatial attention mechanism to highlight important objects. Furthermore, the sensorial attention mechanism extracts location and category information in a pseudo label to guide the LPM for label prediction and avoid details from being ignored. Experimental results on three public HSI datasets show that the proposed method is able to accurately recognize different objects with an overall accuracy (OA) of 98.69%, 98.89%, and 97.56%, respectively.https://www.mdpi.com/2072-4292/16/7/1253hyperspectral image classificationattention mechanismspectral-spatial-sensorial attention networkcontrollable factors |
| spellingShingle | Sheng Li Mingwei Wang Chong Cheng Xianjun Gao Zhiwei Ye Wei Liu Spectral-Spatial-Sensorial Attention Network with Controllable Factors for Hyperspectral Image Classification Remote Sensing hyperspectral image classification attention mechanism spectral-spatial-sensorial attention network controllable factors |
| title | Spectral-Spatial-Sensorial Attention Network with Controllable Factors for Hyperspectral Image Classification |
| title_full | Spectral-Spatial-Sensorial Attention Network with Controllable Factors for Hyperspectral Image Classification |
| title_fullStr | Spectral-Spatial-Sensorial Attention Network with Controllable Factors for Hyperspectral Image Classification |
| title_full_unstemmed | Spectral-Spatial-Sensorial Attention Network with Controllable Factors for Hyperspectral Image Classification |
| title_short | Spectral-Spatial-Sensorial Attention Network with Controllable Factors for Hyperspectral Image Classification |
| title_sort | spectral spatial sensorial attention network with controllable factors for hyperspectral image classification |
| topic | hyperspectral image classification attention mechanism spectral-spatial-sensorial attention network controllable factors |
| url | https://www.mdpi.com/2072-4292/16/7/1253 |
| work_keys_str_mv | AT shengli spectralspatialsensorialattentionnetworkwithcontrollablefactorsforhyperspectralimageclassification AT mingweiwang spectralspatialsensorialattentionnetworkwithcontrollablefactorsforhyperspectralimageclassification AT chongcheng spectralspatialsensorialattentionnetworkwithcontrollablefactorsforhyperspectralimageclassification AT xianjungao spectralspatialsensorialattentionnetworkwithcontrollablefactorsforhyperspectralimageclassification AT zhiweiye spectralspatialsensorialattentionnetworkwithcontrollablefactorsforhyperspectralimageclassification AT weiliu spectralspatialsensorialattentionnetworkwithcontrollablefactorsforhyperspectralimageclassification |