基于图像识别技术的河流流量测验方法研究进展

Research Progress on River Discharge Measurement Method Based on Image Recognition Technology

河流流量监测对洪水灾害防治和河流管理具有重要意义,然而传统的接触式方法逐渐难以满足当下测流需求。视觉流量测验方法是一种利用视频图像识别表面流速和计算流量的非接触式流量测验方法,与传统的接触式方法相比具有高智能和低成本的优点,可以实现无人化和规模化推广应用,对现有水文监测技术提供有力支撑与补充。目前已发展出基于粒子图像识别、水流时空影像、概率和变分以及深度学习的多种图像测速算法,方法精度和适应性得到多场景验证,然而图像采集条件、复杂气象和水域环境以及参数的不确定性仍然制约着这项技术的发展。通过总结视觉测流实施过程存在的问题,并对现有方法提出改进需求,为进一步研究和应用提供借鉴。

River discharge monitoring is important for flood prevention and river management. However, traditional contact methods are increasingly struggling to meet the current requirements. Visual discharge measurement is a non-contact flow test method that uses video images to identify the surface flow velocity and calculate the discharge, which has the advantages of high intelligence and low cost compared with the traditional contact methods. It can achieve unmanned and wide application, and provide a strong support and supplement to the hydrological monitoring technology. At present, there are various image velocimetry algorithms based on particle image recognition, space-time images of water flow, deep learning methods, probability and variational methods, and the accuracy and adaptability of them have been verified in multiple scenarios. However, the image acquisition conditions, complex meteorological and water environments, and parameter uncertainties still constrain the development of this technology. By summarising the problems in the implementation process of vision-based discharge measurement and putting forward the improvement needs for the existing methods, this paper provides a reference for further research and application.