Robust SAR Image Despeckling by Deep Learning From Near-Real Datasets
The inherent speckle in synthetic aperture radar (SAR) images significantly affects their potential usefulness, and its effective suppression is a challenging and nontrivial task. This article uses near-real SAR intensity datasets as the training data for the first time and proposes a robust deep le...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2024-01-01
|
| Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10368288/ |
| _version_ | 1797352854543925248 |
|---|---|
| author | Jianjun Guan Rui Liu Xin Tian Xinming Tang Song Li |
| author_facet | Jianjun Guan Rui Liu Xin Tian Xinming Tang Song Li |
| author_sort | Jianjun Guan |
| collection | DOAJ |
| description | The inherent speckle in synthetic aperture radar (SAR) images significantly affects their potential usefulness, and its effective suppression is a challenging and nontrivial task. This article uses near-real SAR intensity datasets as the training data for the first time and proposes a robust deep learning-based speckle removal model: phase-guided deep despeckling network (PGD2Net). Owing to the unique geometric distortions as well as complicated speckle noise in SAR images it is extremely difficult to simulate geometric distortion similar to real SAR image. In addition, simulating noise as uniform and single-distributed also fails to fully represent speckle complexity. This article uses the temporal and spatial information of time series SAR images to create near-real SAR intensity datasets using an adaptive multilook method called generalized likelihood ratio test, which outstandingly solves the problems encountered with simulated data. Based on the correlation between intensity and phase, to improve the accuracy of speckle noise estimation, we introduce phase information in one subnetwork (speckle noise estimation subnetwork) of the proposed PGD2Net. Our ablation experiments demonstrate that this further enhances the network's speckle suppression performance. Moreover, we establish another subnetwork (dual-branch denoising subnetwork) to conduct feature interaction and estimate the clean intensity image based on a specially designed cross-attention module. Quantitative and qualitative results demonstrate that, compared to other algorithms, our proposed method exhibits strong adaptability and performance across scenes with varying degrees of geometric distortion and speckle due to different terrain undulations. Simultaneously, we extend this algorithm to SAR data obtained from different sensors, achieving excellent execution performance as well. |
| first_indexed | 2024-03-08T13:22:44Z |
| format | Article |
| id | doaj.art-8b39e61192524470a08a70325e6777e5 |
| institution | Directory Open Access Journal |
| issn | 2151-1535 |
| language | English |
| last_indexed | 2024-03-08T13:22:44Z |
| publishDate | 2024-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| spelling | doaj.art-8b39e61192524470a08a70325e6777e52024-01-18T00:00:05ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing2151-15352024-01-01172963297910.1109/JSTARS.2023.334553810368288Robust SAR Image Despeckling by Deep Learning From Near-Real DatasetsJianjun Guan0https://orcid.org/0009-0009-0564-1281Rui Liu1https://orcid.org/0000-0002-0084-150XXin Tian2https://orcid.org/0000-0003-1993-2708Xinming Tang3Song Li4https://orcid.org/0000-0003-2163-8452School of Electronic Information, Wuhan University, Wuhan, ChinaSchool of Electronic Information, Wuhan University, Wuhan, ChinaSchool of Electronic Information, Wuhan University, Wuhan, ChinaLand Satellite Remote Sensing Application Center, Ministry of Natural Resources (MNR), Beijing, ChinaSchool of Electronic Information, Wuhan University, Wuhan, ChinaThe inherent speckle in synthetic aperture radar (SAR) images significantly affects their potential usefulness, and its effective suppression is a challenging and nontrivial task. This article uses near-real SAR intensity datasets as the training data for the first time and proposes a robust deep learning-based speckle removal model: phase-guided deep despeckling network (PGD2Net). Owing to the unique geometric distortions as well as complicated speckle noise in SAR images it is extremely difficult to simulate geometric distortion similar to real SAR image. In addition, simulating noise as uniform and single-distributed also fails to fully represent speckle complexity. This article uses the temporal and spatial information of time series SAR images to create near-real SAR intensity datasets using an adaptive multilook method called generalized likelihood ratio test, which outstandingly solves the problems encountered with simulated data. Based on the correlation between intensity and phase, to improve the accuracy of speckle noise estimation, we introduce phase information in one subnetwork (speckle noise estimation subnetwork) of the proposed PGD2Net. Our ablation experiments demonstrate that this further enhances the network's speckle suppression performance. Moreover, we establish another subnetwork (dual-branch denoising subnetwork) to conduct feature interaction and estimate the clean intensity image based on a specially designed cross-attention module. Quantitative and qualitative results demonstrate that, compared to other algorithms, our proposed method exhibits strong adaptability and performance across scenes with varying degrees of geometric distortion and speckle due to different terrain undulations. Simultaneously, we extend this algorithm to SAR data obtained from different sensors, achieving excellent execution performance as well.https://ieeexplore.ieee.org/document/10368288/DespeckedGeneralized likelihood ratio test (GLRT)near-real synthetic aperture radar (SAR) datasetsphase-guided deep despeckling Network (PGD2Net) |
| spellingShingle | Jianjun Guan Rui Liu Xin Tian Xinming Tang Song Li Robust SAR Image Despeckling by Deep Learning From Near-Real Datasets IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing Despecked Generalized likelihood ratio test (GLRT) near-real synthetic aperture radar (SAR) datasets phase-guided deep despeckling Network (PGD2Net) |
| title | Robust SAR Image Despeckling by Deep Learning From Near-Real Datasets |
| title_full | Robust SAR Image Despeckling by Deep Learning From Near-Real Datasets |
| title_fullStr | Robust SAR Image Despeckling by Deep Learning From Near-Real Datasets |
| title_full_unstemmed | Robust SAR Image Despeckling by Deep Learning From Near-Real Datasets |
| title_short | Robust SAR Image Despeckling by Deep Learning From Near-Real Datasets |
| title_sort | robust sar image despeckling by deep learning from near real datasets |
| topic | Despecked Generalized likelihood ratio test (GLRT) near-real synthetic aperture radar (SAR) datasets phase-guided deep despeckling Network (PGD2Net) |
| url | https://ieeexplore.ieee.org/document/10368288/ |
| work_keys_str_mv | AT jianjunguan robustsarimagedespecklingbydeeplearningfromnearrealdatasets AT ruiliu robustsarimagedespecklingbydeeplearningfromnearrealdatasets AT xintian robustsarimagedespecklingbydeeplearningfromnearrealdatasets AT xinmingtang robustsarimagedespecklingbydeeplearningfromnearrealdatasets AT songli robustsarimagedespecklingbydeeplearningfromnearrealdatasets |