MFAENet: A Multiscale Feature Adaptive Enhancement Network for SAR Image Despeckling
The existence of speckles in synthetic aperture radar (SAR) images affects its subsequent application in computer vision tasks, so the research of speckle suppression plays a very important role. Convolutional neural networks based speckle suppression algorithms cannot reach a good balance between d...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2023-01-01
|
| Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10295981/ |
| _version_ | 1797357973763260416 |
|---|---|
| author | Shuaiqi Liu Luyao Zhang Shikang Tian Qi Hu Bing Li Yudong Zhang |
| author_facet | Shuaiqi Liu Luyao Zhang Shikang Tian Qi Hu Bing Li Yudong Zhang |
| author_sort | Shuaiqi Liu |
| collection | DOAJ |
| description | The existence of speckles in synthetic aperture radar (SAR) images affects its subsequent application in computer vision tasks, so the research of speckle suppression plays a very important role. Convolutional neural networks based speckle suppression algorithms cannot reach a good balance between despeckling effect and structure detail preservation. Considering these issues, a multiscale feature adaptive enhance network for suppressing speckle is proposed. Specifically, an encoder–decoder architecture embedded with multiscale operations is constructed to capture rich contextual information and remove speckles from coarse to fine. Then, deformable convolution is introduced to flexibly adapt changes in ground objects’ complex and diverse image features. Also, the constructed feature adaptive mixup module mitigates shallow feature degradation in deep networks by establishing connections between shallow image texture features and deep image semantic features with learnable weights. Experiments on synthetic and real SAR images show that the proposed method produces advanced results regarding visual quality and objective metrics. |
| first_indexed | 2024-03-08T14:53:25Z |
| format | Article |
| id | doaj.art-83ddc96c075a41c18d8237c7ff734621 |
| institution | Directory Open Access Journal |
| issn | 2151-1535 |
| language | English |
| last_indexed | 2024-03-08T14:53:25Z |
| publishDate | 2023-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| spelling | doaj.art-83ddc96c075a41c18d8237c7ff7346212024-01-11T00:01:02ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing2151-15352023-01-0116104201043310.1109/JSTARS.2023.332733210295981MFAENet: A Multiscale Feature Adaptive Enhancement Network for SAR Image DespecklingShuaiqi Liu0https://orcid.org/0000-0001-7520-8226Luyao Zhang1https://orcid.org/0000-0003-4009-4657Shikang Tian2https://orcid.org/0009-0005-6178-8393Qi Hu3https://orcid.org/0000-0002-7495-1066Bing Li4https://orcid.org/0000-0002-5888-6735Yudong Zhang5https://orcid.org/0000-0002-4870-1493College of Electronic and Information Engineering, Hebei University, Machine Vision Technology Innovation Center of Hebei Province, Baoding, ChinaCollege of Electronic and Information Engineering, Hebei University, Key Laboratory of Digital Medical Engineering of Hebei Province, Baoding, ChinaCollege of Electronic and Information Engineering, Hebei University, Key Laboratory of Digital Medical Engineering of Hebei Province, Baoding, ChinaInstitute of Information Science, Beijing Jiaotong University, Beijing, ChinaState Key Laboratory of Multimodal Artificial Intelligence Systems, Institute of Automation, Chinese Academy of Sciences, Beijing, ChinaSchool of Computing and Mathematics, University of Leicester, Leicester, U.K.The existence of speckles in synthetic aperture radar (SAR) images affects its subsequent application in computer vision tasks, so the research of speckle suppression plays a very important role. Convolutional neural networks based speckle suppression algorithms cannot reach a good balance between despeckling effect and structure detail preservation. Considering these issues, a multiscale feature adaptive enhance network for suppressing speckle is proposed. Specifically, an encoder–decoder architecture embedded with multiscale operations is constructed to capture rich contextual information and remove speckles from coarse to fine. Then, deformable convolution is introduced to flexibly adapt changes in ground objects’ complex and diverse image features. Also, the constructed feature adaptive mixup module mitigates shallow feature degradation in deep networks by establishing connections between shallow image texture features and deep image semantic features with learnable weights. Experiments on synthetic and real SAR images show that the proposed method produces advanced results regarding visual quality and objective metrics.https://ieeexplore.ieee.org/document/10295981/Adaptive fusionfeature enhancementmultiscale featurespeckle suppressionsynthetic aperture radar (SAR) images |
| spellingShingle | Shuaiqi Liu Luyao Zhang Shikang Tian Qi Hu Bing Li Yudong Zhang MFAENet: A Multiscale Feature Adaptive Enhancement Network for SAR Image Despeckling IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing Adaptive fusion feature enhancement multiscale feature speckle suppression synthetic aperture radar (SAR) images |
| title | MFAENet: A Multiscale Feature Adaptive Enhancement Network for SAR Image Despeckling |
| title_full | MFAENet: A Multiscale Feature Adaptive Enhancement Network for SAR Image Despeckling |
| title_fullStr | MFAENet: A Multiscale Feature Adaptive Enhancement Network for SAR Image Despeckling |
| title_full_unstemmed | MFAENet: A Multiscale Feature Adaptive Enhancement Network for SAR Image Despeckling |
| title_short | MFAENet: A Multiscale Feature Adaptive Enhancement Network for SAR Image Despeckling |
| title_sort | mfaenet a multiscale feature adaptive enhancement network for sar image despeckling |
| topic | Adaptive fusion feature enhancement multiscale feature speckle suppression synthetic aperture radar (SAR) images |
| url | https://ieeexplore.ieee.org/document/10295981/ |
| work_keys_str_mv | AT shuaiqiliu mfaenetamultiscalefeatureadaptiveenhancementnetworkforsarimagedespeckling AT luyaozhang mfaenetamultiscalefeatureadaptiveenhancementnetworkforsarimagedespeckling AT shikangtian mfaenetamultiscalefeatureadaptiveenhancementnetworkforsarimagedespeckling AT qihu mfaenetamultiscalefeatureadaptiveenhancementnetworkforsarimagedespeckling AT bingli mfaenetamultiscalefeatureadaptiveenhancementnetworkforsarimagedespeckling AT yudongzhang mfaenetamultiscalefeatureadaptiveenhancementnetworkforsarimagedespeckling |