Spatial-Adaptive Siamese Residual Network for Multi-/Hyperspectral Classification
Deep learning methods have been successfully applied for multispectral and hyperspectral images classification due to their ability to extract hierarchical abstract features. However, the performance of these methods relies heavily on large-scale training samples. In this paper, we propose a three-d...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2020-05-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/12/10/1640 |
| _version_ | 1827716511951749120 |
|---|---|
| author | Zhi He Dan He |
| author_facet | Zhi He Dan He |
| author_sort | Zhi He |
| collection | DOAJ |
| description | Deep learning methods have been successfully applied for multispectral and hyperspectral images classification due to their ability to extract hierarchical abstract features. However, the performance of these methods relies heavily on large-scale training samples. In this paper, we propose a three-dimensional spatial-adaptive Siamese residual network (3D-SaSiResNet) that requires fewer samples and still enhances the performance. The proposed method consists of two main steps: construction of 3D spatial-adaptive patches and Siamese residual network for multiband images classification. In the first step, the spectral dimension of the original multiband images is reduced by a stacked autoencoder and superpixels of each band are obtained by the simple linear iterative clustering (SLIC) method. Superpixels of the original multiband image can be finally generated by majority voting. Subsequently, the 3D spatial-adaptive patch of each pixel is extracted from the original multiband image by reference to the previously generated superpixels. In the second step, a Siamese network composed of two 3D residual networks is designed to extract discriminative features for classification and we train the 3D-SaSiResNet by pairwise inputting the training samples into the networks. The testing samples are then fed into the trained 3D-SaSiResNet and the learned features of the testing samples are classified by the nearest neighbor classifier. Experimental results on three multiband image datasets show the feasibility of the proposed method in enhancing classification performance even with limited training samples. |
| first_indexed | 2024-03-10T19:42:41Z |
| format | Article |
| id | doaj.art-4ad024cd2ca44f0dbfe2ff16389b88dd |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-10T19:42:41Z |
| publishDate | 2020-05-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-4ad024cd2ca44f0dbfe2ff16389b88dd2023-11-20T01:07:48ZengMDPI AGRemote Sensing2072-42922020-05-011210164010.3390/rs12101640Spatial-Adaptive Siamese Residual Network for Multi-/Hyperspectral ClassificationZhi He0Dan He1Guangdong Provincial Key Laboratory of Urbanization and Geo-Simulation, Center of Integrated Geographic Information Analysis, Southern Marine Science and Engineering Guangdong Laboratory (Zhuhai), School of Geography and Planning, Sun Yat-sen University, Guangzhou 510275, ChinaCity College of Dongguan University of Technology, Dongguan 511700, ChinaDeep learning methods have been successfully applied for multispectral and hyperspectral images classification due to their ability to extract hierarchical abstract features. However, the performance of these methods relies heavily on large-scale training samples. In this paper, we propose a three-dimensional spatial-adaptive Siamese residual network (3D-SaSiResNet) that requires fewer samples and still enhances the performance. The proposed method consists of two main steps: construction of 3D spatial-adaptive patches and Siamese residual network for multiband images classification. In the first step, the spectral dimension of the original multiband images is reduced by a stacked autoencoder and superpixels of each band are obtained by the simple linear iterative clustering (SLIC) method. Superpixels of the original multiband image can be finally generated by majority voting. Subsequently, the 3D spatial-adaptive patch of each pixel is extracted from the original multiband image by reference to the previously generated superpixels. In the second step, a Siamese network composed of two 3D residual networks is designed to extract discriminative features for classification and we train the 3D-SaSiResNet by pairwise inputting the training samples into the networks. The testing samples are then fed into the trained 3D-SaSiResNet and the learned features of the testing samples are classified by the nearest neighbor classifier. Experimental results on three multiband image datasets show the feasibility of the proposed method in enhancing classification performance even with limited training samples.https://www.mdpi.com/2072-4292/12/10/1640remote sensingclassificationstacked autoencodersuperpixelSiamese network |
| spellingShingle | Zhi He Dan He Spatial-Adaptive Siamese Residual Network for Multi-/Hyperspectral Classification Remote Sensing remote sensing classification stacked autoencoder superpixel Siamese network |
| title | Spatial-Adaptive Siamese Residual Network for Multi-/Hyperspectral Classification |
| title_full | Spatial-Adaptive Siamese Residual Network for Multi-/Hyperspectral Classification |
| title_fullStr | Spatial-Adaptive Siamese Residual Network for Multi-/Hyperspectral Classification |
| title_full_unstemmed | Spatial-Adaptive Siamese Residual Network for Multi-/Hyperspectral Classification |
| title_short | Spatial-Adaptive Siamese Residual Network for Multi-/Hyperspectral Classification |
| title_sort | spatial adaptive siamese residual network for multi hyperspectral classification |
| topic | remote sensing classification stacked autoencoder superpixel Siamese network |
| url | https://www.mdpi.com/2072-4292/12/10/1640 |
| work_keys_str_mv | AT zhihe spatialadaptivesiameseresidualnetworkformultihyperspectralclassification AT danhe spatialadaptivesiameseresidualnetworkformultihyperspectralclassification |