A sample generation and enhancement method for side-scan sonar targets in complex marine environments

doi:10.11947/j.AGCS.2026.20250503

Abstract

Abstract:

To address the difficulty of acquiring real side-scan sonar (SSS) target samples and the limited capability of conventional augmentation methods in representing acoustic scattering characteristics, this study proposes a target sample generation framework based on the SSS imaging mechanism. An underwater target imaging model integrating echo intensity and acoustic ray propagation is first constructed according to the SSS imaging principle and acoustic energy propagation model. Subsequently, considering the influence of seabed topography and geomorphology on acoustic imaging, an improved style transfer strategy is introduced to enhance the generated images and improve their realism and environmental consistency. In addition, an environmental noise model is incorporated to perform noise perturbation and environmental feature enhancement, thereby generating multi-scene shipwreck target samples. Experimental results demonstrate that the detection model trained solely on the generated samples achieves a detection accuracy of 0.88 mAP (0.5), verifying the effectiveness of the proposed sample generation method for few-shot underwater target detection tasks. The proposed method can provide data support for training side-scan sonar target detection models in complex marine environments.

Key words: side-scan sonar, sample amplification, style transfer, underwater object detection, deep learning

CLC Number:

P228

Xi Zhao, Qiangqiang Yuan, Jiadan Xu. A sample generation and enhancement method for side-scan sonar targets in complex marine environments[J]. Acta Geodaetica et Cartographica Sinica, 2026, 55(7): 1229-1239.

Add to citation manager EndNote|Reference Manager|ProCite|BibTeX|RefWorks

URL: http://xb.chinasmp.com/EN/10.11947/j.AGCS.2026.20250503

http://xb.chinasmp.com/EN/Y2026/V55/I7/1229

Figures/Tables 9

Fig. 1

Fig. 2

Fig. 3

Fig. 4

Fig. 5

Fig. 6

Tab. 1

Tab. 2

Tab. 3

References 30

[1]	杨元喜, 徐天河, 薛树强. 我国海洋大地测量基准与海洋导航技术研究进展与展望[J]. 测绘学报, 2017, 46(1): 1-8. DOI: . doi: 10.11947/j.AGCS.2017.20160519
	Yang Yuanxi, Xu Tianhe, Xue Shuqiang. Progresses and prospects in developing marine geodetic datum and marine navigation of China[J]. Acta Geodaetica et Cartographica Sinica, 2017, 46(1): 1-8. DOI: . doi: 10.11947/j.AGCS.2017.20160519
[2]	赵建虎, 欧阳永忠, 王爱学. 海底地形测量技术现状及发展趋势[J]. 测绘学报, 2017, 46(10): 1786-1794. DOI: . doi: 10.11947/j.AGCS.2017.20170276
	Zhao Jianhu, Ouyang Yongzhong, Wang Aixue. Status and development tendency for seafloor terrain measurement technology[J]. Acta Geodaetica et Cartographica Sinica, 2017, 46(10): 1786-1794. DOI: . doi: 10.11947/j.AGCS.2017.20170276
[3]	Zhao Xi, Yuan Qiangqiang, Xu Jiadan. Efficient processing of side-scan sonar images and fast detection of sparse targets in large-scale images[J]. International Journal of Digital Earth, 2025, 18(1): 2510568.
[4]	王晓. 侧扫声呐图像精处理及目标识别方法研究[J]. 测绘学报, 2021, 50(2): 282. DOI: . doi: 10.11947/j.AGCS.2021.20190517
	Wang Xiao. Research on precise processing of side scan sonar image and object recognition methods[J]. Acta Geodaetica et Cartographica Sinica, 2021, 50(2): 282. DOI: . doi: 10.11947/j.AGCS.2021.20190517
[5]	赵建虎, 王爱学, 王晓, 等. 侧扫声纳条带图像分段拼接方法研究[J]. 武汉大学学报(信息科学版), 2013, 38(9): 1034-1038.
	Zhao Jianhu, Wang Aixue, Wang Xiao, et al. A segmented mosaic method for side scan sonar strip images using corresponding features[J]. Geomatics and Information Science of Wuhan University, 2013, 38(9): 1034-1038.
[6]	于永灿, 赵建虎, 李冰墨, 等. 侧扫声呐图像目标识别模型类增量更新方法[J]. 测绘学报, 2026, 55(3): 502-514. DOI: . doi: 10.11947/j.AGCS.2026.20250362
	Yu Yongcan, Zhao Jianhu, Li Bingmo, et al. Class-incremental update method for target recognition models in side-scan sonar images[J]. Acta Geodaetica et Cartographica Sinica, 2026, 55(3): 502-514. DOI: . doi: 10.11947/j.AGCS.2026.20250362
[7]	汤寓麟, 金绍华, 边刚, 等. 侧扫声呐识别沉船影像的迁移学习卷积神经网络法[J]. 测绘学报, 2021, 50(2): 260-269. DOI: . doi: 10.11947/j.AGCS.2021.20200187
	Tang Yulin, Jin Shaohua, Bian Gang, et al. The transfer learning with convolutional neural network method of side-scan sonar to identify wreck images[J]. Acta Geodaetica et Cartographica Sinica, 2021, 50(2): 260-269. DOI: . doi: 10.11947/j.AGCS.2021.20200187
[8]	胡涛. 水下目标可视化仿真技术研究[D]. 哈尔滨: 哈尔滨工程大学, 2007.
	Hu Tao. The study on visual simulation techniques of underwater target[D]. Harbin: Harbin Engineering University, 2007.
[9]	龚权华, 朱维强, 夏显文. 基于样本扩增的水下集装箱智能识别[J]. 海洋测绘, 2022, 42(4): 22-26.
	Gong Quanhua, Zhu Weiqiang, Xia Xianwen. An intelligent recognition of underwater containers based on sample augmentation[J]. Hydrographic Surveying and Charting, 2022, 42(4): 22-26.
[10]	Tang Yulin, Wang Liming, Bian Shaofeng, et al. SSS underwater target image samples augmentation based on the cross-domain mapping relationship of images of the same physical object[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2023, 16: 6393-6410.
[11]	Huang Chao, Zhao Jianhu, Yu Yongcan, et al. Comprehensive sample augmentation by fully considering SSS imaging mechanism and environment for shipwreck detection under zero real samples[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 5906814.
[12]	Chen Yuwei, Pei S C. Domain adaptation for underwater image enhancement via content and style separation[J]. IEEE Access, 2022, 10: 90523-90534.
[13]	Shorten C, Khoshgoftaar T M. A survey on image data augmentation for deep learning[J]. Journal of Big Data, 2019, 6(1): 60.
[14]	Huo Guanying, Wu Ziyin, Li Jiabiao. Underwater object classification in sidescan sonar images using deep transfer learning and semisynthetic training data[J]. IEEE Access, 2020, 8: 47407-47418.
[15]	Wu Yanchen, Wang Yingmin. A research on underwater target recognition neural network for small samples[J]. Journal of Northwestern Polytechnical University, 2022, 40(1): 40-46.
[16]	Gong Jianya, Ji Shunping. Photogrammetry and deep learning[J]. Journal of Geodesy and Geoinformation Science, 2018, 1(1): 1-15.
[17]	Chen L C, Papandreou G, Schroff F, et al. Rethinking atrous convolution for semantic image segmentation[EB/OL]. (2017-12-05) [2025-11-12]. https://doi.org/10.48550/arXiv.1706.05587.
[18]	Qian Xiaoqi, Liu Weifeng, Zhang Jing, et al. Underwater-relevant image object detection based feature-degraded enhancement method[J]. Journal of Image and Graphics, 2022, 27(11): 3185-3198.
[19]	Wei Chengwei, Bai Yunfei, Liu Chang, et al. Unsupervised underwater shipwreck detection in side-scan sonar images based on domain-adaptive techniques[J]. Scientific Reports, 2024, 14: 12687.
[20]	Sung M S, Kim J, Lee M, et al. Realistic sonar image simulation using deep learning for underwater object detection[J]. International Journal of Control, Automation and Systems, 2020, 18(3): 523-534.
[21]	Li Shaobo, Ma Jinfeng, Wu Yunlong, et al. SSS small target detection via combining weighted sparse model with shadow characteristics[J]. IEEE Transactions on Geoscience and Remote Sensing, 2023, 61: 5911711.
[22]	Sung M S, Song Y W, Yu S C. Underwater object detection of AUV based on sonar simulator utilizing noise addition[C]//Proceedings of 2022 IEEE/OES Autonomous Underwater Vehicles Symposium. Singapore: IEEE, 2022: 1-5.
[23]	Deng Yingying, Tang Fan, Dong Weiming, et al. StyTr2: image style transfer with transformers[C]//Proceedings of 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New Orleans: IEEE, 2022.
[24]	Key W H. Side scan sonar technology[C]//Proceedings of OCEANS 2000 MTS/IEEE Conference and Exhibition, 2000: 1029-1033.
[25]	Li Shaobo, Zhao Jianhu, Yu Yongcan, et al. Anisotropic total variation regularized low-rank approximation for SSS images radiometric distortion correction[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 5925412.
[26]	Daniel S, Le Leannec F, Roux C, et al. Side-scan sonar image matching[J]. IEEE Journal of Oceanic Engineering, 1998, 23(3): 245-259.
[27]	Ainslie M A. The sonar equation and the definitions of propagation loss[J]. The Journal of the Acoustical Society of America, 2004, 115(1): 131-134.
[28]	Luo Yang, Wu Aiping, Fu Qingqing. MAS-YOLOv11: an improved underwater object detection algorithm based on YOLOv11[J]. Sensors, 2025, 25(11): 3433.
[29]	Robinson S P, Lepper P A, Hazelwood R A. Good practice guide for underwater noise measurement: 133[R]. National Physical Laboratory, 2014.
[30]	Jong C A F D. Characterization of ships as sources of underwater noise[C]//Proceedings of 2009 International Conference on Acoustics. Rotterdam: [s.n.], 2009.

数据集类型	样本	训练集	数据集样本量
测试集	实测SSS样本	Te	101
训练集	实测SSS样本	Tr1	328
	基于Tr1进行旋转、缩放操作扩增后的SSS样本	Tr1^*	854
	原始仿真SSS样本	Tr2	427
	仿真目标与背景叠加后样本	Tr3	854
	Tr3基础上加噪后的仿真SSS样本	Tr4	854
	Tr3基础上风格迁移后的仿真SSS样本	Tr5	854
	Tr3基础上先加噪后风格迁移的仿真样本	Tr6	854
	Tr3基础上先风格迁移后加噪的仿真样本	Tr7	854
	部分Tr7仿真样本（526张）与Tr1实测样本（328张）合集	Tr7^*	854
	全部仿真样本（Tr2—Tr7）	Tr8	4697

模型	训练集	P	R	mAP（0.5）	mAP（0.5～0.95）	平均推理时间/（ms/frame）
1	Tr1	0.870	0.879	0.893	0.636	9.382
1^*	Tr1^*	0.931	0.876	0.917	0.646	9.412
2	Tr2	0.537	0.664	0.576	0.301	9.523
3	Tr3	0.719	0.542	0.642	0.373	9.471
4	Tr4	0.811	0.8	0.805	0.486	9.44
5	Tr5	0.748	0.738	0.757	0.434	9.493
6	Tr6	0.779	0.759	0.774	0.531	9.434
7	Tr7	0.879	0.869	0.880	0.623	9.467
7^*	Tr7^*	0.890	0.881	0.902	0.640	9.501
8	Tr8	0.908	0.888	0.909	0.643	9.655

模型	真实样本数	样本数仿真	总数	仿真样本比例/（%）	mAP（0.5）
Mix-A	554	300	854	35.1	0.892
Mix-B（基线模型）	328	526	854	61.6	0.902
Mix-C	154	700	854	82.0	0.910