Chen Meng, Chen Hua, Wu Zeheng, Huang Yu, Zhou Nie, Xu Chong-Yu
State Key Laboratory of Water Resources Engineering and Management, Wuhan University, Wuhan 430072, China.
Department of Geosciences, University of Oslo, N-0316 Oslo, Norway.
Sensors (Basel). 2024 Jul 18;24(14):4655. doi: 10.3390/s24144655.
The hydrological monitoring of flow data is important for flood prevention and modern river management. However, traditional contact methods are increasingly struggling to meet the requirements of simplicity, accuracy, and continuity. The video-based river discharge measurement is a technique to monitor flow velocity without contacting the water body by using the image-recognition algorithms, which has been verified to have the advantages of full coverage and full automation compared with the traditional contact technique. In order to provide a timely summary of the available results and to inform further research and applications, this paper reviews and synthesizes the literature on the general implementation routes of the video-based river discharge measurement technique and the principles and advances of today's popular image-recognition algorithms for velocity detection. Then, it discusses the challenges of image-recognition algorithms in terms of image acquisition conditions, parameter uncertainties, and complex meteorological and water environments. It is concluded that the performance of this technique can be improved by enhancing the robustness and accuracy of video-based discharge measurement algorithms, minimizing weather effects, and improving computational efficiency. Finally, future development directions for further perfecting this technique are outlined.
流量数据的水文监测对于防洪和现代河流管理至关重要。然而，传统的接触式方法越来越难以满足简单性、准确性和连续性的要求。基于视频的河流流量测量是一种利用图像识别算法在不接触水体的情况下监测流速的技术，与传统接触式技术相比，已被证实具有全覆盖和全自动化的优势。为了及时总结现有成果并为进一步的研究和应用提供参考，本文回顾并综合了关于基于视频的河流流量测量技术的一般实施路线以及当今流行的速度检测图像识别算法的原理和进展的文献。然后，讨论了图像识别算法在图像采集条件、参数不确定性以及复杂气象和水环境方面面临的挑战。得出的结论是，可以通过提高基于视频的流量测量算法的鲁棒性和准确性、最小化天气影响以及提高计算效率来提升该技术的性能。最后，概述了进一步完善该技术的未来发展方向。
