Hyperspectral Image Classification Based on Multiscale Hybrid Networks and Attention Mechanisms
Hyperspectral image (HSI) classification is one of the most crucial tasks in remote sensing processing. The attention mechanism is preferable to a convolutional neural network (CNN), due to its superior ability to express information during HSI processing. Recently, numerous methods combining CNNs a...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-05-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/15/11/2720 |
| _version_ | 1797596844474236928 |
|---|---|
| author | Haizhu Pan Xiaoyu Zhao Haimiao Ge Moqi Liu Cuiping Shi |
| author_facet | Haizhu Pan Xiaoyu Zhao Haimiao Ge Moqi Liu Cuiping Shi |
| author_sort | Haizhu Pan |
| collection | DOAJ |
| description | Hyperspectral image (HSI) classification is one of the most crucial tasks in remote sensing processing. The attention mechanism is preferable to a convolutional neural network (CNN), due to its superior ability to express information during HSI processing. Recently, numerous methods combining CNNs and attention mechanisms have been applied in HSI classification. However, it remains a challenge to achieve high-accuracy classification by fully extracting effective features from HSIs under the conditions of limited labeled samples. In this paper, we design a novel HSI classification network based on multiscale hybrid networks and attention mechanisms. The network consists of three subnetworks: a spectral-spatial feature extraction network, a spatial inverted pyramid network, and a classification network, which are employed to extract spectral-spatial features, to extract spatial features, and to obtain classification results, respectively. The multiscale fusion network and attention mechanisms complement each other by capturing local and global features separately. In the spatial pyramid network, multiscale spaces are formed through down-sampling, which can reduce redundant information while retaining important information. The structure helps the network better capture spatial features at different scales, and to improve classification accuracy. Experimental results on various public HSI datasets demonstrate that the designed network is extremely competitive compared to current advanced approaches, under the condition of insufficient samples. |
| first_indexed | 2024-03-11T02:58:45Z |
| format | Article |
| id | doaj.art-97f2a7a5d8ac4c00bb001410bc8fe761 |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-11T02:58:45Z |
| publishDate | 2023-05-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-97f2a7a5d8ac4c00bb001410bc8fe7612023-11-18T08:27:47ZengMDPI AGRemote Sensing2072-42922023-05-011511272010.3390/rs15112720Hyperspectral Image Classification Based on Multiscale Hybrid Networks and Attention MechanismsHaizhu Pan0Xiaoyu Zhao1Haimiao Ge2Moqi Liu3Cuiping Shi4College of Computer and Control Engineering, Qiqihar University, Qiqihar 161000, ChinaCollege of Computer and Control Engineering, Qiqihar University, Qiqihar 161000, ChinaCollege of Computer and Control Engineering, Qiqihar University, Qiqihar 161000, ChinaCollege of Computer and Control Engineering, Qiqihar University, Qiqihar 161000, ChinaCollege of Telecommunication and Electronic Engineering, Qiqihar University, Qiqihar 161000, ChinaHyperspectral image (HSI) classification is one of the most crucial tasks in remote sensing processing. The attention mechanism is preferable to a convolutional neural network (CNN), due to its superior ability to express information during HSI processing. Recently, numerous methods combining CNNs and attention mechanisms have been applied in HSI classification. However, it remains a challenge to achieve high-accuracy classification by fully extracting effective features from HSIs under the conditions of limited labeled samples. In this paper, we design a novel HSI classification network based on multiscale hybrid networks and attention mechanisms. The network consists of three subnetworks: a spectral-spatial feature extraction network, a spatial inverted pyramid network, and a classification network, which are employed to extract spectral-spatial features, to extract spatial features, and to obtain classification results, respectively. The multiscale fusion network and attention mechanisms complement each other by capturing local and global features separately. In the spatial pyramid network, multiscale spaces are formed through down-sampling, which can reduce redundant information while retaining important information. The structure helps the network better capture spatial features at different scales, and to improve classification accuracy. Experimental results on various public HSI datasets demonstrate that the designed network is extremely competitive compared to current advanced approaches, under the condition of insufficient samples.https://www.mdpi.com/2072-4292/15/11/2720hyperspectral image classificationmultiscale hybrid networkhybrid attention mechanismmulti-head attention mechanism |
| spellingShingle | Haizhu Pan Xiaoyu Zhao Haimiao Ge Moqi Liu Cuiping Shi Hyperspectral Image Classification Based on Multiscale Hybrid Networks and Attention Mechanisms Remote Sensing hyperspectral image classification multiscale hybrid network hybrid attention mechanism multi-head attention mechanism |
| title | Hyperspectral Image Classification Based on Multiscale Hybrid Networks and Attention Mechanisms |
| title_full | Hyperspectral Image Classification Based on Multiscale Hybrid Networks and Attention Mechanisms |
| title_fullStr | Hyperspectral Image Classification Based on Multiscale Hybrid Networks and Attention Mechanisms |
| title_full_unstemmed | Hyperspectral Image Classification Based on Multiscale Hybrid Networks and Attention Mechanisms |
| title_short | Hyperspectral Image Classification Based on Multiscale Hybrid Networks and Attention Mechanisms |
| title_sort | hyperspectral image classification based on multiscale hybrid networks and attention mechanisms |
| topic | hyperspectral image classification multiscale hybrid network hybrid attention mechanism multi-head attention mechanism |
| url | https://www.mdpi.com/2072-4292/15/11/2720 |
| work_keys_str_mv | AT haizhupan hyperspectralimageclassificationbasedonmultiscalehybridnetworksandattentionmechanisms AT xiaoyuzhao hyperspectralimageclassificationbasedonmultiscalehybridnetworksandattentionmechanisms AT haimiaoge hyperspectralimageclassificationbasedonmultiscalehybridnetworksandattentionmechanisms AT moqiliu hyperspectralimageclassificationbasedonmultiscalehybridnetworksandattentionmechanisms AT cuipingshi hyperspectralimageclassificationbasedonmultiscalehybridnetworksandattentionmechanisms |