Swin-ReshoUnet: A Seismic Profile Signal Reconstruction Method Integrating Hierarchical Convolution, ORCA Attention, and Residual Channel Attention Mechanism
This study proposes a Swin-ReshoUnet architecture with a three-level enhancement mechanism to address inefficiencies in multi-scale feature extraction and gradient degradation in deep networks for high-precision seismic exploration. The encoder uses a hierarchical convolution module to build a multi...
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-07-01
|
| Series: | Applied Sciences |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2076-3417/15/15/8332 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849770679430283264 |
|---|---|
| author | Jie Rao Mingju Chen Xiaofei Song Chen Xie Xueyang Duan Xiao Hu Senyuan Li Xingyue Zhang |
| author_facet | Jie Rao Mingju Chen Xiaofei Song Chen Xie Xueyang Duan Xiao Hu Senyuan Li Xingyue Zhang |
| author_sort | Jie Rao |
| collection | DOAJ |
| description | This study proposes a Swin-ReshoUnet architecture with a three-level enhancement mechanism to address inefficiencies in multi-scale feature extraction and gradient degradation in deep networks for high-precision seismic exploration. The encoder uses a hierarchical convolution module to build a multi-scale feature pyramid, enhancing cross-scale geological signal representation. The decoder replaces traditional self-attention with ORCA attention to enable global context modeling with lower computational cost. Skip connections integrate a residual channel attention module, mitigating gradient degradation via dual-pooling feature fusion and activation optimization, forming a full-link optimization from low-level feature enhancement to high-level semantic integration. Simulated and real dataset experiments show that at decimation ratios of 0.1–0.5, the method significantly outperforms SwinUnet, TransUnet, etc., in reconstruction performance. Residual signals and F-K spectra verify high-fidelity reconstruction. Despite increased difficulty with higher sparsity, it maintains optimal performance with notable margins, demonstrating strong robustness. The proposed hierarchical feature enhancement and cross-scale attention strategies offer an efficient seismic profile signal reconstruction solution and show generality for migration to complex visual tasks, advancing geophysics-computer vision interdisciplinary innovation. |
| format | Article |
| id | doaj-art-3365609f35394ba598a89fca10ea1f0c |
| institution | DOAJ |
| issn | 2076-3417 |
| language | English |
| publishDate | 2025-07-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Applied Sciences |
| spelling | doaj-art-3365609f35394ba598a89fca10ea1f0c2025-08-20T03:02:55ZengMDPI AGApplied Sciences2076-34172025-07-011515833210.3390/app15158332Swin-ReshoUnet: A Seismic Profile Signal Reconstruction Method Integrating Hierarchical Convolution, ORCA Attention, and Residual Channel Attention MechanismJie Rao0Mingju Chen1Xiaofei Song2Chen Xie3Xueyang Duan4Xiao Hu5Senyuan Li6Xingyue Zhang7College of Automation and Information Engineering, Sichuan University of Science and Engineering, Yibin 644002, ChinaCollege of Automation and Information Engineering, Sichuan University of Science and Engineering, Yibin 644002, ChinaCollege of Automation and Information Engineering, Sichuan University of Science and Engineering, Yibin 644002, ChinaCollege of Automation and Information Engineering, Sichuan University of Science and Engineering, Yibin 644002, ChinaCollege of Automation and Information Engineering, Sichuan University of Science and Engineering, Yibin 644002, ChinaCollege of Automation and Information Engineering, Sichuan University of Science and Engineering, Yibin 644002, ChinaCollege of Automation and Information Engineering, Sichuan University of Science and Engineering, Yibin 644002, ChinaCollege of Automation and Information Engineering, Sichuan University of Science and Engineering, Yibin 644002, ChinaThis study proposes a Swin-ReshoUnet architecture with a three-level enhancement mechanism to address inefficiencies in multi-scale feature extraction and gradient degradation in deep networks for high-precision seismic exploration. The encoder uses a hierarchical convolution module to build a multi-scale feature pyramid, enhancing cross-scale geological signal representation. The decoder replaces traditional self-attention with ORCA attention to enable global context modeling with lower computational cost. Skip connections integrate a residual channel attention module, mitigating gradient degradation via dual-pooling feature fusion and activation optimization, forming a full-link optimization from low-level feature enhancement to high-level semantic integration. Simulated and real dataset experiments show that at decimation ratios of 0.1–0.5, the method significantly outperforms SwinUnet, TransUnet, etc., in reconstruction performance. Residual signals and F-K spectra verify high-fidelity reconstruction. Despite increased difficulty with higher sparsity, it maintains optimal performance with notable margins, demonstrating strong robustness. The proposed hierarchical feature enhancement and cross-scale attention strategies offer an efficient seismic profile signal reconstruction solution and show generality for migration to complex visual tasks, advancing geophysics-computer vision interdisciplinary innovation.https://www.mdpi.com/2076-3417/15/15/8332Swin-ReshoUnetthree-level enhancementhierarchical convolutionORCA attentionresidual channel attentionseismic profile signal reconstruction |
| spellingShingle | Jie Rao Mingju Chen Xiaofei Song Chen Xie Xueyang Duan Xiao Hu Senyuan Li Xingyue Zhang Swin-ReshoUnet: A Seismic Profile Signal Reconstruction Method Integrating Hierarchical Convolution, ORCA Attention, and Residual Channel Attention Mechanism Applied Sciences Swin-ReshoUnet three-level enhancement hierarchical convolution ORCA attention residual channel attention seismic profile signal reconstruction |
| title | Swin-ReshoUnet: A Seismic Profile Signal Reconstruction Method Integrating Hierarchical Convolution, ORCA Attention, and Residual Channel Attention Mechanism |
| title_full | Swin-ReshoUnet: A Seismic Profile Signal Reconstruction Method Integrating Hierarchical Convolution, ORCA Attention, and Residual Channel Attention Mechanism |
| title_fullStr | Swin-ReshoUnet: A Seismic Profile Signal Reconstruction Method Integrating Hierarchical Convolution, ORCA Attention, and Residual Channel Attention Mechanism |
| title_full_unstemmed | Swin-ReshoUnet: A Seismic Profile Signal Reconstruction Method Integrating Hierarchical Convolution, ORCA Attention, and Residual Channel Attention Mechanism |
| title_short | Swin-ReshoUnet: A Seismic Profile Signal Reconstruction Method Integrating Hierarchical Convolution, ORCA Attention, and Residual Channel Attention Mechanism |
| title_sort | swin reshounet a seismic profile signal reconstruction method integrating hierarchical convolution orca attention and residual channel attention mechanism |
| topic | Swin-ReshoUnet three-level enhancement hierarchical convolution ORCA attention residual channel attention seismic profile signal reconstruction |
| url | https://www.mdpi.com/2076-3417/15/15/8332 |
| work_keys_str_mv | AT jierao swinreshounetaseismicprofilesignalreconstructionmethodintegratinghierarchicalconvolutionorcaattentionandresidualchannelattentionmechanism AT mingjuchen swinreshounetaseismicprofilesignalreconstructionmethodintegratinghierarchicalconvolutionorcaattentionandresidualchannelattentionmechanism AT xiaofeisong swinreshounetaseismicprofilesignalreconstructionmethodintegratinghierarchicalconvolutionorcaattentionandresidualchannelattentionmechanism AT chenxie swinreshounetaseismicprofilesignalreconstructionmethodintegratinghierarchicalconvolutionorcaattentionandresidualchannelattentionmechanism AT xueyangduan swinreshounetaseismicprofilesignalreconstructionmethodintegratinghierarchicalconvolutionorcaattentionandresidualchannelattentionmechanism AT xiaohu swinreshounetaseismicprofilesignalreconstructionmethodintegratinghierarchicalconvolutionorcaattentionandresidualchannelattentionmechanism AT senyuanli swinreshounetaseismicprofilesignalreconstructionmethodintegratinghierarchicalconvolutionorcaattentionandresidualchannelattentionmechanism AT xingyuezhang swinreshounetaseismicprofilesignalreconstructionmethodintegratinghierarchicalconvolutionorcaattentionandresidualchannelattentionmechanism |