Article Details

Abstract

To address the large fluctuation in feature point number and insufficient tracking stability of visual-inertial SLAM front-end caused by significant illumination variation and strong light scattering in underwater environments, this paper proposes an improved visual-inertial SLAM front-end method tailored for underwater applications. The proposed method uses the number of feature points as a feedback metric, establishes a scene feature baseline, and dynamically adjusts the parameters of contrast-limited adaptive histogram equalization to achieve adaptive enhancement under complex underwater lighting conditions. In addition, an improved ORB feature extraction strategy combined with a bidirectional optical flow consistency check is employed to reduce mismatched features induced by underwater noise and image degradation. Experiments conducted on the MCUVI underwater visual-inertial dataset demonstrate that the proposed method improves front-end feature tracking stability under various underwater illumination conditions, and achieves higher absolute trajectory accuracy and improved robustness compared with the original VINS-Mono on sequences where the system runs stably.

Keywords

underwater visual-inertial SLAM; feature point feedback; feature tracking

References

1. [1] Liu Haomin, Zhang Guofeng, Bao Hujun Overview of Simultaneous Localization and Map Construction Methods Based on Monocular Vision [J]. Journal of Computer Aided Design and Graphics, 2016, 28 (6): 855-868
2. [2] Mourikis A I，Roumeliotis S I. A Multi-State Constraint Kalman Filter for Vision-aided Inertial Navigation [C]//Proceedings 2007 IEEE International Conference on Robotics and Automation，2007：3565-3572.
3. [3] Leutenegger S，Lynen S，Bosse M，et al. Keyframe-based Visual–Inertial Odometry Using Nonlinear Optimization [J].The International Journal of Robotics Research，2015，34（3）：314-334.
4. [4] Qin T，Li P L，Shen S J. VINS-Mono: A Robust and Versatile Monocular Visual-Inertial State Estimator [J].IEEE Transactions on Robotics，2018，34（4）：1004-1020.
5. [5] Mur-Artal R，Montiel J M M，Tardós J D. ORB-SLAM： A Versatile and Accurate Monocular SLAM System [J].IEEE Transactions on Robotics，2015，31（5）：1147-1163.
6. [6] Song J B，Li W Q，Zhu X W. Acoustic-VINS: Tightly Coupled Acoustic-Visual-Inertial Navigation System for Autonomous Underwater Vehicles [J].IEEE Robotics and Automation Letters，2024，9（2）：1620-1627.
7. [7] Xu Z Z，Haroutunian M，Murphy A J，et al. An Integrated Visual Odometry System With Stereo Camera for Unmanned Underwater Vehicles [J].IEEE Access，2022，10：71329-71343.
8. [8] Ou Y M，Fan J F，Zhou C，et al. Hybrid-VINS: Underwater Tightly Coupled Hybrid Visual Inertial Dense SLAM for AUV [J].IEEE Transactions on Industrial Electronics，2025，72（3）：2821-28315.
9. [9] Zhang B B，Liu S，Ji D X，et al. MVINS: A Magnetism&vision Aided Inertial Navigation System for Autonomous Underwater Vehicles [J].IEEE Robotics and Automation Letters，2025，10（3）：2239-2246.
10. [10] Wu R H，Gao Y Z. Research on Underwater Complex Scene SLAM Algorithm Based on Image Enhancement [J/OL].Sensors，2022，22（21）：8517.https://doi.org/10.3390/s22218517.
11. [11] Wang Haifang, Qiao Dingjie, Wu Tianhao, etc State estimation method based on improved VINS Mono algorithm [J]. Journal of Northeastern University (Natural Science Edition), 2024, 45 (9): 1287-1293
12. [12] Li Zhizheng, Nie Zhixi, Wang Zhenjie, etc Improving the visual/inertial fusion positioning algorithm for visual front-end [J]. Global Positioning System, 2024, 49 (3): 20-27
13. [13] Pizer S M，Johnston R E，Ericksen J P，et al. Contrast-limited adaptive histogram equalization: speed and effectiveness [C]//Proceedings of the First Conference on Visualization in Biomedical Computing，1990：337-345.
14. [14] Zhang Meng, Tang Qingling, Jiang Xiaofei Adaptive AGAST feature homogenization extraction algorithm [J]. Intelligent Computers and Applications, 2023, 13 (8): 66-72
15. [15] Yang Hongfan, Li Hang, Chen Kaiyang, etc Image feature point extraction and matching method based on improved ORB algorithm [J]. Journal of Cartography, 2020, 41 (4): 548-555
16. [16] Rosten E，Drummond T. Machine Learning for High-Speed Corner Detection [C]//Computer Vision-ECCV 2006.ECCV 2006. Lecture Notes in Computer Science，2006：430-443.
17. [17] Yu Yongjie, Shen Yichun, Zhou Zhifeng, etc Synchronized positioning and mapping algorithm based on improved VINS [J]. Journal of Weapon Equipment Engineering, 2024, 45 (7): 283-289297
18. [18] Singh M，Dharmadhikari M，Alexis K. An Online Self-calibrating Refractive Camera Model with Application to Underwater Odometry [C]//2024 IEEE International Conference on Robotics and Automation(ICRA)，2024：10005-10011.

About the authors

Cao Xincheng (1992-), male, Han ethnicity, from Qingzhou, Shandong Province, lecturer, doctoral student, research direction: design and development of marine intelligent equipment; Zheng Shihao (2001-), male, Han ethnicity, from Ningde, Fujian, is currently pursuing a master's degree in underwater positioning and navigation.