Underwater Target Detection Based on Sonar Image
-
摘要: 通过处理声呐图像实现水下目标检测具有重大的军事与民用意义。文章对基于声呐图像的水下目标检测原理、方法、算法和发展趋势进行了全面阐述。 将基于声呐图像的水下目标检测任务分为传统水下目标检测、基于深度学习的目标检测, 以及深度学习与迁移学习相结合的目标检测3个层面。又分别将传统目标检测分为基于数理统计、基于数学形态学以及基于像素的水下目标检测; 将基于深度学习的目标检测分为基于一阶段方法、基于二阶段方法以及基于DETR的目标检测; 将深度学习与迁移学习相结合的目标检测分为基于简单深度神经网络模型的迁移与基于复杂深度学习模型的迁移所实现的目标检测进行了具体讨论。最后总结归纳现有技术的优缺点, 并对该领域的未来发展方向作出进一步的展望。Abstract: Underwater target detection by processing sonar images is of great military and civil significance. This paper comprehensively describes the principles, methods, algorithms, and development trends in underwater target detection based on sonar images. Initially, we divide the underwater target detection task based on sonar images into traditional, deep learning-based, and combined deep learning- and transfer learning-based underwater target detection. Traditional target detection is divided into underwater target detection based on mathematical statistics, mathematical morphology, and pixels. Deep learning-based target detection methods are primarily divided into one-stage, two-stage, and detection transformer(DETR) methods. Combined deep learning- and transfer learning-based target detection is primarily divided into target detection based on simple deep neural network model transfer and complex deep learning model transfer. Finally, the advantages and disadvantages of the existing technology are summarized, and the future development direction of this field is discussed.
-
Key words:
- underwater target detection /
- sonar image /
- deep learning /
- transfer learning
-
表 1 多种成像声呐类型及优缺点
Table 1. Advantages and disadvantages of various imaging sonars
成像声呐类型 优点 缺点 侧扫声呐 扫描分辨率高; 经济高效 存在声脉冲散射问题 合成孔径声呐 分辨力高; 抗干扰能力强 硬件要求高 多波束声呐 探测距离远; 精度高 操作难度高 前视声呐 低功耗; 实用价值高 存在高旁瓣现象 
下载: 导出CSV
表 2 不同纹理特征的对比
Table 2. Comparison of texture features
纹理特征 优点 缺点 灰度共生矩阵 描述具有方向性及灰度差异大的纹理图像时效果好 计算量大; 特征量之间的统计相关性普遍存在 Tamura纹理 具有良好的旋转不变性与尺度不变性 描述声呐图像的局部纹理特征时效果差 分形纹理特征 描述整体纹理与局部纹理的效果均良好; 识别效率高 精确率提升小
下载: 导出CSV
表 3 应用于声呐图像水下目标检测的数学形态学方法对比
Table 3. Comparison of mathematical morphology methods applied to sonar image in underwater target detection
数学形态
学方法优点 缺点 中值滤波 部署简便; 速度快 抑制噪声效果较差 Lee滤波 无需精确建模; 能较好保持边缘 计算量大; 速度较慢 Kuan滤波 滤除散斑噪声、平滑图像时对边缘不产生影响 边缘保持能力差 Frost滤波 抑制噪声效果较好 边缘细节等结构信息模糊; 存在盲目平滑现象
下载: 导出CSV
表 4 YOLO 与SSD的对比
Table 4. Comparison between YOLO and SSD
SSD YOLO 损失函数 Softmax loss Logistic loss 特征提取器 VGG19 Darknet-53 包围框预测 默认框直接偏移 网格单元通过Sigmoid激活函数偏移 对小目标的检测效果 底层的语义值不高, 对小目标的检测效果差 分辨率较高的层有较高的语义值, 对小目标的检测较好 对大目标的检测效果 较好 较差
下载: 导出CSV
-
[1] 贾宇. 关于海洋强国战略的思考[J]. 太平洋学报, 2018, 26(1): 1-8. doi: 10.14015/j.cnki.1004-8049.2018.01.001Jia Yu. On China’s maritime power strategy[J]. Pacific Journal, 2018, 26(1): 1-8. doi: 10.14015/j.cnki.1004-8049.2018.01.001 [2] Bryner D, Huffer F, Srivastava A, et al. Underwater minefield detection in clutter data using spatial point-process models[J]. IEEE Journal of Oceanic Engineering, 2016, 41(3): 670-681. doi: 10.1109/JOE.2015.2493598 [3] Ceccarini C. The multi-role nato-interoperable light weight torpedo for the 21st century[J]. Information & Security an International Journal, 2004, 13(1): 89-97. [4] Chuang M C, Hwang J N, Ye J H, et al. Underwater fish tracking for moving cameras based on deformable multiple kernels[J]. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 2016, 47(9): 1-11. [5] 张韫峰, 李娟, 黎明. 基于图像处理的水下海参识别和定位方法[J]. 水下无人系统学报, 2021, 29(1): 111-123. doi: 10.11993/j.issn.2096-3920.2021.01.016Zhang Yunfeng, Li Juan, Li Ming. Underwater sea cucumber identification and localization method based on image processing[J]. Journal of Unmanned Undersea Systems, 2021, 29(1): 111-123. doi: 10.11993/j.issn.2096-3920.2021.01.016 [6] Byun S W, Kim J Y. Development and experiment of a hovering AUV tested-bed for underwater exploration[C]// Symposium on Underwater Technology & Workshop on Scientific Use of Submarine Cables& Related Technologies. Tokyo, Japan: IEEE, 2007. [7] 李勇航, 牟泽霖, 万芃. 海洋侧扫声呐探测技术的现状及发展[J]. 通讯世界, 2015(3): 213-214. doi: 10.3969/j.issn.1006-4222.2015.03.137 [8] Hayes M P, Gough P T. Synthetic aperture sonar: A review of current status[J]. IEEE Journal of Oceanic Engineering, 2009, 34(3): 207-224. doi: 10.1109/JOE.2009.2020853 [9] Komatsu T, Igarashi C, Tatsukawa K, et al. Use of multi-beam sonar to map seagrass beds in otsuchi bay on the Sanriku Coast of Japan[J]. Aquatic Living Resources, 2003, 16(3): 223-230. doi: 10.1016/S0990-7440(03)00045-7 [10] Lowe D G. Distinctive image features from scale-invariant key points[J]. International Journal of Computer Vision, 2003, 20: 91-110. [11] Bay H, Tuytelaars T, Gool L V. SURF: Speeded up robust features[C]//Proceedings of the 9th European Conference on Computer Vision-Volume Part I. Berlin, Heidelberg: Springer-Verlag, 2006. [12] Dalal N, Triggs B. Histograms of oriented gradients for human detection[C]//IEEE Computer Society Conference on Computer Vision & Pattern Recognition. San Diego, CA, USA: IEEE, 2005. [13] Burges C. A Tutorial on support vector machines for pattern recognition[J]. Data Mining and Knowledge Discovery, 1998, 2(2): 121-167. doi: 10.1023/A:1009715923555 [14] Rish I. An empirical study of the naive Bayes classifier[J]. Journal of Universal Computer Science, 2001, 1(2): 127. [15] Krizhevsky A, Sutskever I, Hinton G. ImageNet classification with deep convolutional neural networks[J]. Communications of the ACM, 2017, 60: 84-90. [16] Lin T Y, Dollar P, Girshick R, et al. Feature pyramid networks for object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Honolulu, HI, USA: IEEE, 2017: 2117-2125. [17] 陈强, 田杰, 黄海宁, 等. 基于统计和纹理特征的SAS图像SVM分割研究[J]. 仪器仪表学报, 2013, 34(6): 214-221. doi: 10.3969/j.issn.0254-3087.2013.06.031Chen Qiang, Tian Jie, Huang Haining, et al. Study on SAS image segmentation using SVM based on statistical and texture features[J]. Chinese Journal of Scientific Instrument, 2013, 34(6): 214-221. doi: 10.3969/j.issn.0254-3087.2013.06.031 [18] 王涛, 潘国富, 张济博. 基于侧扫声呐图像纹理特征的海底底质分类研究[C]//2020中国西部声学学术交流会论文集. 酒泉: 2020中国西部声学学术交流会, 2020: 403-406. [19] 董凌宇, 单瑞, 刘慧敏, 等. 基于分形纹理特征的侧扫声呐图像沉船识别方法研究[J]. 海洋地质与第四纪地质, 2021, 41(4): 232-239. doi: 10.16562/j.cnki.0256-1492.2020070301Dong Lingyu, Shan Rui, Liu Huimin, et al. Shipwreck identification with side scan sonar image based on fractal texture[J]. Marine Geology & Quaternary Geology, 2021, 41(4): 232-239. doi: 10.16562/j.cnki.0256-1492.2020070301 [20] 王其林, 王宏健, 李庆, 等. 侧扫声呐图像特征提取改进方法[J]. 水下无人系统学报, 2019, 27(3): 297-304.Wang Qilin, Wang Hongjian, Li Qing, et al. Improved feature extraction method for side scan sonar images[J]. Journal of Unmanned Undersea Systems, 2019, 27(3): 297-304. [21] 田晓东, 刘忠, 周德超. 基于形状描述直方图的声呐图像目标识别算法[J]. 系统工程与电子技术, 2007(7): 1049-1052,1212. doi: 10.3321/j.issn:1001-506X.2007.07.008Tian Xiaodong, Liu Zhong, Zhou Dechao. Mine target recognition algorithm of sonar image[J]. Systems Engineering and Electronics, 2007(7): 1049-1052,1212. doi: 10.3321/j.issn:1001-506X.2007.07.008 [22] 卢逢春, 张殿伦, 郭海涛. 基于属性直方图的图像分割方法及其在声呐图像分割中的应用[J]. 哈尔滨工程大学学报, 2002(3): 1-3,19. doi: 10.3969/j.issn.1006-7043.2002.03.001Lu Fengchun, Zhang Dianlun, Guo Haitao. Image segmentation based upon bounded histogram and its application to sonar image segmentation[J]. Journal of Harbin Engineering University, 2002(3): 1-3,19. doi: 10.3969/j.issn.1006-7043.2002.03.001 [23] Yang F, Du Z, Wu Z. Object recognizing on sonar image based on histogram and geometric feature[J]. Marine Science Bulletin, 2006, 25(5): 64. [24] 王晓, 邹泽伟, 李勃勃, 等. 基于多特征融合的彩色图像声呐目标检测[J]. 计算机科学, 2019, 46(S1): 177-181. [25] Sun C, Wang L, Wang N, et al. Image recognition technology in texture identification of marine sediment sonar image[J]. Complexity, 2021, 2021(2): 1-8. [26] 罗进华, 蒋锦朋, 朱培民. 基于数学形态学的侧扫声呐图像轮廓自动提取[J]. 海洋学报, 2016, 38(5): 150-157.Luo Jinhua, Jiang Jinpeng, Zhu Peimin. Automatic extraction of the side-scan sonar imagery outlines based on mathematical morphology[J]. Haiyang Xuebao, 2016, 38(5): 150-157. [27] 邹岗, 田晓东, 刘忠. 基于几何特征的声呐图像人造目标检测算法[J]. 舰船科学技术, 2007, 29(6): 177-179, 183.Zou Gang, Tian Xiaodong, Liu Zhong. Man-made target detection algorithm of sonar image based on geometric feature[J]. Ship Science and Technology, 2007, 29(6): 177-179, 183. [28] Mallet J, Courmontagne P. A new wavelet thresholding approach for SAS images denoising[C]//Oceans, 2006. Boston, USA: IEEE, 2006: 1-6. [29] Isar A, Isar D, Moga S, et al. Multi-scale MAP despeckling of sonar images[C]//Oceans 2005-Europe. Brest, France: IEEE, 2005. [30] 金凤来, 钟何平. 一种改进的合成孔径声呐图像Lee滤波算法[J]. 舰船电子工程, 2017, 37(3): 35-37. doi: 10.3969/j.issn.1672-9730.2017.03.009Jin Fenglai, Zhong Heping. An improved Lee filter algorithm for synthetic aperture sonar[J]. Ship Electronic Engineering, 2017, 37(3): 35-37. doi: 10.3969/j.issn.1672-9730.2017.03.009 [31] 郭海涛, 徐雷, 赵红叶, 等. 一种抑制声呐图像散斑噪声的形态学滤波器[J]. 仪器仪表学报, 2015, 36(3): 654-660. doi: 10.19650/j.cnki.cjsi.2015.03.022Guo Haitao, Xu Lei, Zhao Hongye, et al. A morphological filter for despeckling of a sonar image[J]. Chinese Journal of Scientific Instrument, 2015, 36(3): 654-660. doi: 10.19650/j.cnki.cjsi.2015.03.022 [32] 崔杰, 胡长青, 徐海东. 基于帧差法的多波束前视声呐运动目标检测[J]. 仪器仪表学报, 2018, 39(2): 169-176. doi: 10.19650/j.cnki.cjsi.j1702414Cui Jie, Hu Changqing, Xu Haidong. Moving target detection for multi-beam forward-looking sonar based on frame-difference method[J]. Chinese Journal of Scientific Instrument, 2018, 39(2): 169-176. doi: 10.19650/j.cnki.cjsi.j1702414 [33] 崔杰, 胡长青, 徐海东, 等. 改进的Mean Shift算法在前视声呐运动目标跟踪中的应用[J]. 声学技术, 2020, 39(3): 279-283.Cui Jie, Hu Changqing, Xu Haidong, et al. Application of improved mean shift algorithm in moving target tracking of forward-looking sonar[J]. Technical Acoustics, 2020, 39(3): 279-283. [34] Liu H, Dai J, Wang R, et al. Combining background subtraction and three-frame difference to detect moving object from underwater video[C]//Oceans-shanghai. Shanghai, China: IEEE, 2016. [35] Kalyan B, Balasuriya A. Sonar based automatic target detection scheme for underwater environments using CFAR techniques: A comparative study[C]//Proceedings of the 2004 International Symposium on Underwater Technology(IEEE Cat. No. 04EX869). Taipei: IEEE, 2005. [36] Li K, Liu Z, Hu S L, et al. Target detection of sonar image based on bis-parameter with adaptive windows[J]. Applied Mechanics & Materials, 2013, 347-350: 3407-3410. [37] Villar S A, Acosta G G, Solari F J. OS-CFAR process in 2-D for object segmentation from side-scan sonar data[C]//Information Processing & Control. Cordoba, Argentina: IEEE, 2016. [38] Girshick R, Donahue J, Darrell T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Columbus, USA: IEEE, 2014: 580-587. [39] Girshick R. Fast R-CNN[C]//Proceedings of the IEEE International Conference on Computer Vision. Santiago, Chile: IEEE, 2015: 1440-1448. [40] Ren S, He K, Girshick R, et al. Faster R-CNN: Towards real-time object detection with region proposal net- works[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2017, 39(6): 1137-1149. [41] 曾文冠, 鲁建华. 基于卷积神经网络的声呐图像目标检测识别[C]//第十七届船舶水下噪声学术讨论会论文集. 衢州, 中国: 《船舶力学》编辑部, 2019. [42] Fang J, Wang P. Target detection in sonar image based on Faster RCNN[C]//2020 International Conference on Information Science, Parallel and Distributed Systems(ISPDS). Xi’an, China: IEEE, 2020. [43] Ma Q, Jiang L, Yu W, et al. Training with noise adversarial network: A generalization method for object detection on sonar image[C]//Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. Snowmass, USA: IEEE, 2020: 729-738. [44] 马麒翔. 基于深度学习的声呐图像目标检测算法研究[D]. 南京: 东南大学, 2020. [45] Goodfellow I J, Pouget-Abadie J, Mirza M, et al. Generative adversarial nets[C]//NIPS’14: Proceedings of the 27th International Conference on Neural Information Processing Systems. [S.l.]: MIT Press, 2014. [46] Redmon J, Divvala S, Girshick R, et al. You only look once: unified, real-time object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas, USA: IEEE, 2016: 779-788. [47] Liu W, Anguelov D, Erhan D, et al. SSD: Single shot multibox detector[C]//Computer Vision ECCV 2016. European Conference on Computer Vision-Amsterdam. Cham: Springer, 2016, 9905: 21-37. [48] Wu Y. Sonar image target detection and recognition based on convolution neural network[J]. Mobile Information Systems, 2021, 2021(6): 1-8. [49] 王霞. 面向声呐图像的水下小目标识别方法研究[D]. 哈尔滨: 哈尔滨工程大学, 2020. [50] Fan X, Lu L, Shi P, et al. A novel sonar target detection and classification algorithm[J]. Multimedia Tools and Applications, 2022, 81(7): 10091-10106. doi: 10.1007/s11042-022-12054-4 [51] Han K, Wang Y, Chen H, et al. A survey on vision transformer[C]//IEEE Transactions on Pattern Analysis and Machine Intelligence. Piscataway, USA: IEEE, 2022. [52] Yu Y, Zhao J, Gong Q, et al. Real-time underwater maritime object detection in side-scan sonar images based on transformer-YOLOV5[J]. Remote Sensing, 2021, 13(18): 3555. doi: 10.3390/rs13183555 [53] 凡志邈, 李海林, 夏伟杰, 等. 基于深度学习的成像声纳水下目标的检测与分类[C]//中国声学学会水声学分会2019年学术会议论文集. 南京, 中国: 中国声学学会水声学分会, 2019. [54] 李宝奇, 黄海宁, 刘纪元, 等. 基于改进SSD的合成孔径声呐图像水下多尺度目标轻量化检测模型[J]. 电子与信息学报, 2021, 43(10): 2854-62. doi: 10.11999/JEIT201042Li Baoqi, Huang Haining, Liu Jiyuan, et al. Synthetic aperture sonar underwater multi-scale target efficient detection model based on improved single shot detector[J]. Journal of Electronics & Information Technology, 2021, 43(10): 2854-62. doi: 10.11999/JEIT201042 [55] Carion N, Massa F, Synnaeve G, et al. End-to-end object detection with transformers[C]//European Conference on Computer Vision. Cham: Springer, 2020: 213-229. [56] Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need[J]. Advances in Neural Information Processing Systems, 2017, 30: 5998-6008. [57] 汤寓麟, 李厚朴, 张卫东, 等. 侧扫声纳检测沉船目标的轻量化DETR-YOLO法[J]. 系统工程与电子技术, 2022, 44(8): 2427-2436. doi: 10.12305/j.issn.1001-506X.2022.08.06Tang Yulin, Li Houpu, Zhang Weidong, et al. Lightweight DETR-YOLO method for detecting shipwreck target in side-scan sonar[J]. Systems Engineering and Electronics, 2022, 44(8): 2427-2436. doi: 10.12305/j.issn.1001-506X.2022.08.06 [58] Patel V M, Gopalan R, Li R, et al. Visual domain adaptation: A survey of recent advances[J]. Signal Processing Magazine, IEEE, 2015, 32(3): 53-69. doi: 10.1109/MSP.2014.2347059 [59] 朱兆彤, 付学志, 胡友峰. 一种利用迁移学习训练卷积神经网络的声呐图像识别方法[J]. 水下无人系统学报, 2020, 28(1): 89-96. doi: 10.11993/j.issn.2096-3920.2020.01.013Zhu Zhaotong, Fu Xuezhi, Hu Youfeng. A sonar image recognition method based on convolutional neural network trained through transfer learning[J]. Journal of Unmanned Undersea Systems, 2020, 28(1): 89-96. doi: 10.11993/j.issn.2096-3920.2020.01.013 [60] 武铄, 王晓, 张丹阳, 等. 联合迁移学习和深度学习的侧扫声呐沉船识别方法[J]. 河南科技, 2021, 40(36): 36-40. doi: 10.3969/j.issn.1003-5168.2021.36.015 [61] 于淼. 基于深度学习的侧扫声呐图像目标检测方法研究[D]. 哈尔滨: 哈尔滨工程大学, 2020. [62] 盛子旗, 霍冠英. 样本仿真结合迁移学习的声呐图像水雷检测[J]. 智能系统学报, 2021, 16(2): 385-392. doi: 10.11992/tis.202101030Sheng Ziqi, Huo Guanying. Detection of underwater mine target in side scan sonar image based on sample simulation and transfer learning[J]. CAAI Transactions on Intelligent Systems, 2021, 16(2): 385-392. doi: 10.11992/tis.202101030 [63] Tang Y, Jin S, Bian G, et al. Shipwreck target recognition in side-scan sonar images by improved YOLOv3 model based on transfer learning[J]. IEEE Access, 2020, 8: 173450-173460. doi: 10.1109/ACCESS.2020.3024813 [64] 付同强, 胡桥, 刘钰, 等. 基于优化二维变分模态分解与迁移学习的水下目标识别方法[J]. 水下无人系统学报, 2021, 29(2): 153-163. doi: 10.11993/j.issn.2096-3920.2021.02.004Fu Tongqiang, Hu Qiao, Liu Yu, et al. Underwater target identification method based on optimized 2D variational mode decomposition and transfer learning[J]. Journal of Unmanned Undersea Systems, 2021, 29(2): 153-163. doi: 10.11993/j.issn.2096-3920.2021.02.004 -
点击查看大图
图(3) / 表(4)
计量
- 文章访问数: 307
- HTML全文浏览量: 19
- PDF下载量: 130
- 被引次数: 0
