SonarNet: Global Feature-Based Hybrid Attention Network for Side-Scan Sonar Image Segmentation
With the rapid advancement of deep learning techniques, side-scan sonar image segmentation has become a crucial task in underwater scene understanding. However, the complex and variable underwater environment poses significant challenges for salient object detection, with traditional deep learning a...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-07-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/17/14/2450 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850071504044163072 |
|---|---|
| author | Juan Lei Huigang Wang Liming Fan Qingyue Gu Shaowei Rong Huaxia Zhang |
| author_facet | Juan Lei Huigang Wang Liming Fan Qingyue Gu Shaowei Rong Huaxia Zhang |
| author_sort | Juan Lei |
| collection | DOAJ |
| description | With the rapid advancement of deep learning techniques, side-scan sonar image segmentation has become a crucial task in underwater scene understanding. However, the complex and variable underwater environment poses significant challenges for salient object detection, with traditional deep learning approaches often suffering from inadequate feature representation and the loss of global context during downsampling, thus compromising the segmentation accuracy of fine structures. To address these issues, we propose SonarNet, a Global Feature-Based Hybrid Attention Network specifically designed for side-scan sonar image segmentation. SonarNet features a dual-encoder architecture that leverages residual blocks and a self-attention mechanism to simultaneously capture both global structural and local contextual information. In addition, an adaptive hybrid attention module is introduced to effectively integrate channel and spatial features, while a global enhancement block fuses multi-scale global and spatial representations from the dual encoders, mitigating information loss throughout the network. Comprehensive experiments on a dedicated underwater sonar dataset demonstrate that SonarNet outperforms ten state-of-the-art saliency detection methods, achieving a mean absolute error as low as 2.35%. These results highlight the superior performance of SonarNet in challenging sonar image segmentation tasks. |
| format | Article |
| id | doaj-art-baada02554f148e092ca46efe0880c2d |
| institution | DOAJ |
| issn | 2072-4292 |
| language | English |
| publishDate | 2025-07-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj-art-baada02554f148e092ca46efe0880c2d2025-08-20T02:47:17ZengMDPI AGRemote Sensing2072-42922025-07-011714245010.3390/rs17142450SonarNet: Global Feature-Based Hybrid Attention Network for Side-Scan Sonar Image SegmentationJuan Lei0Huigang Wang1Liming Fan2Qingyue Gu3Shaowei Rong4Huaxia Zhang5School of Marine Science and Technology, Northwestern Polytechnical University, West Youyi Road, Xi’an 710072, ChinaSchool of Marine Science and Technology, Northwestern Polytechnical University, West Youyi Road, Xi’an 710072, ChinaSchool of Marine Science and Technology, Northwestern Polytechnical University, West Youyi Road, Xi’an 710072, ChinaSchool of Equipment Management and UAV of Air Force Engineering University, No. 1 Jiazi, Changle East Road, Xi’an 710045, ChinaSchool of Faculty of Information Engineering and Automation, Kunming University of Science and Technology, Jingming South Road, Kunming 650500, ChinaSchool of Information Science and Engineering, Shandong Agricultural University, Daizong Road, Tai’an 271000, ChinaWith the rapid advancement of deep learning techniques, side-scan sonar image segmentation has become a crucial task in underwater scene understanding. However, the complex and variable underwater environment poses significant challenges for salient object detection, with traditional deep learning approaches often suffering from inadequate feature representation and the loss of global context during downsampling, thus compromising the segmentation accuracy of fine structures. To address these issues, we propose SonarNet, a Global Feature-Based Hybrid Attention Network specifically designed for side-scan sonar image segmentation. SonarNet features a dual-encoder architecture that leverages residual blocks and a self-attention mechanism to simultaneously capture both global structural and local contextual information. In addition, an adaptive hybrid attention module is introduced to effectively integrate channel and spatial features, while a global enhancement block fuses multi-scale global and spatial representations from the dual encoders, mitigating information loss throughout the network. Comprehensive experiments on a dedicated underwater sonar dataset demonstrate that SonarNet outperforms ten state-of-the-art saliency detection methods, achieving a mean absolute error as low as 2.35%. These results highlight the superior performance of SonarNet in challenging sonar image segmentation tasks.https://www.mdpi.com/2072-4292/17/14/2450salient object detectionglobal feature extractionself-attention mechanismunderwater sonarmambahybrid attention mechanism |
| spellingShingle | Juan Lei Huigang Wang Liming Fan Qingyue Gu Shaowei Rong Huaxia Zhang SonarNet: Global Feature-Based Hybrid Attention Network for Side-Scan Sonar Image Segmentation Remote Sensing salient object detection global feature extraction self-attention mechanism underwater sonar mamba hybrid attention mechanism |
| title | SonarNet: Global Feature-Based Hybrid Attention Network for Side-Scan Sonar Image Segmentation |
| title_full | SonarNet: Global Feature-Based Hybrid Attention Network for Side-Scan Sonar Image Segmentation |
| title_fullStr | SonarNet: Global Feature-Based Hybrid Attention Network for Side-Scan Sonar Image Segmentation |
| title_full_unstemmed | SonarNet: Global Feature-Based Hybrid Attention Network for Side-Scan Sonar Image Segmentation |
| title_short | SonarNet: Global Feature-Based Hybrid Attention Network for Side-Scan Sonar Image Segmentation |
| title_sort | sonarnet global feature based hybrid attention network for side scan sonar image segmentation |
| topic | salient object detection global feature extraction self-attention mechanism underwater sonar mamba hybrid attention mechanism |
| url | https://www.mdpi.com/2072-4292/17/14/2450 |
| work_keys_str_mv | AT juanlei sonarnetglobalfeaturebasedhybridattentionnetworkforsidescansonarimagesegmentation AT huigangwang sonarnetglobalfeaturebasedhybridattentionnetworkforsidescansonarimagesegmentation AT limingfan sonarnetglobalfeaturebasedhybridattentionnetworkforsidescansonarimagesegmentation AT qingyuegu sonarnetglobalfeaturebasedhybridattentionnetworkforsidescansonarimagesegmentation AT shaoweirong sonarnetglobalfeaturebasedhybridattentionnetworkforsidescansonarimagesegmentation AT huaxiazhang sonarnetglobalfeaturebasedhybridattentionnetworkforsidescansonarimagesegmentation |