针对水下组合导航特种工况传感器精度不足、数据失效、惯导累积误差等问题，提出基于运动模糊图像的辅助测速方法，该方法在分析水下密封装置及水体折射对成像影响的基础上，建立载体运动速度与图像运动模糊尺度的数学模型；利用潜器本体搭载的视觉系统，获取对底运动的全局模糊图像；通过分析模糊图像的频谱特征，确定频谱图像的暗条纹间距与运动模糊尺度之间的关系；通过角度修正的hough变换直线提取方法得到了频谱图像的暗条纹间距，实现了潜器对底运动速度的估计；水池实验结果测试显示基于水下运动模糊图像测速方法的估算值与实测值的误差小于0.3%，方法可行有效，可应用于无人潜水器在近底作业过程中的辅助测速及定位导航。
To address issues such as insufficient sensor accuracy, data failure, and inertial navigation system (INS) cumulative errors in underwater integrated navigation under special working conditions, this study proposes an auxiliary velocity measurement method based on motion-blurred images. The method establishes a mathematical model between vehicle motion velocity and image motion blur scale by analyzing the effects of underwater sealing devices and water refraction on imaging. Utilizing the visual system mounted on the submersible, global motion-blurred images of seabed movement are acquired. The relationship between dark fringe spacing in spectral images and motion blur scale is determined through analysis of spectral characteristics in blurred images. The method of Hough transform line extraction with angle correction is used to obtain the spacing between dark stripes in the spectrogram, enabling the estimation of the submersible's bottom movement speed. Pool test results demonstrate that the error between estimated velocity derived from underwater motion-blurred images and measured values is less than 0.3%, confirming the feasibility and effectiveness of the method. This approach can be applied to auxiliary velocity measurement and positioning navigation for unmanned underwater vehicles during near-seabed operations.
2026,48(1): 133-140 收稿日期：2025-3-28
DOI：10.3404/j.issn.1672-7649.2026.01.019
分类号：U653；TP249
基金项目：国家自然科学基金资助项目（52231011）；上海交通大学大学生创新实践计划资助项目（IPP29027）
作者简介：刘宗霖(1994-),男,博士研究生,研究方向为水下机器人、组合导航定位、视觉辅助导航
参考文献：
[1] 陈柯宇, 黄南宇, 苏为, 等. 一种基于SINS与LBL的水下隐蔽导航定位方法[J]. 水下无人系统学报, 2022, 30(1): 37-43.
CHEN K Y, HUANG N Y, SU W, et al. Underwater concealed navigation and positioning method based on SINS and LBL[J]. Journal of Unmanned Undersea Systems, 2022, 30(1): 37-43.
[2] WANG Q, LIU K, CAO Z. System noise variance matrix adaptive Kalman filter method for AUV INS/DVL navigation system[J]. Ocean Engineering, 2023, 267: 113269.
[3] LI P, LIU Y, YAN T, et al. A robust INS/USBL/DVL integrated navigation algorithm using graph optimization[J]. Sensors, 2023, 23(2): 916.
[4] FERRERA, MORAS M, TROUVÉ-PELOUX J, et al. Real-time monocular visual odometry for turbid and dynamic underwater environments[J]. Sensors, 2019, 19(3), 687.
[5] 张强, 游子昂, 王建, 等. 水下机器人海底地形主动同步定位与建图具身规划算法[J/OL]. 哈尔滨工程大学学报, 1-7[2025-03-10].
ZHANG Q, YOU Z A, WANG J, et al. Embodiment planning algorithm for AUV bathymetric active SLAM [J]. Journal of Harbin Engineering University, 2025, 1(7): 1-8.
[6] BURGUERA A, BONIN-FONT F, FONT EG, et al. Combining deep learning and robust estimation for outlier-resilient Underwater visual graph SLAM[J]. Journal of Marine Science and Engineering. 2022; 10(4): 511.
[7] AMARASINGHE C, RATNAWEERA A, MAITRIPALA S. UW Deep SLAM-CNN Assisted Underwater SLAM. Appl. Comput. Syst., 2023, 28(1), 100-113.
[8] WANG X, FAN X, SHI P, et al. An overview of key SLAM technologies for underwater scenes[J]. Remote Sensing. 2023; 15(10): 2496.
[9] 朱益贤, 马腾, 范佳佳, 等. 融合测深/测距信息的水下同步定位与建图技术[J]. 无人系统技术, 2023, 6(4): 10-21.
ZHU Y X, MA T, FAN J J, et al. Underwater simultaneous localization and mapping fusing bathymetric and range information[J]. Unmanned Systems Technology, 2023, 6(4): 10-21.
[10] MERVEILLE FFR, JIA B, XU Z, et al. Enhancing underwater SLAM navigation and perception: a comprehensive review of deep learning integration[J]. Sensors, 2024, 24(21): 7034.
[11] 金守峰, 范荻, 田明锐. 基于运动模糊图像的车载视觉测速方法[J]. 中国工程机械学报, 2017, 15(1): 83-89.
JIN S F, FAN D, TIAN M R. Research on vehicle speed measurement method based on motion blur image[J]. Chinese Journal of Construction Machinery, 2017, 15(1): 83-89.
