干渠及河流流速测量是水资源管理及洪涝预测的重要依据。围绕基于机器视觉的干渠流速测量技术和方法进行分析和综合，阐述了其算法的一般框架及各部分功能，重点综述了近年来粒子图像测速法、粒子跟踪测速法、时空图像测速法、光流法，以及基于深度学习的测流方法的原理、技术和发展情况，最后给出了目前尚存在的问题和挑战，指明未来可能的重点发展方向。

The velocity measurement of trunk canals and rivers is regarded as an important basis for water resources management and flood prediction. The techniques and methods for trunk canals velocity measurement based on machine vision were analyzed and synthesized. Particular focus was placed on reviewing the principles, technologies, and recent developments of particle image velocimetry, particle tracking velocimetry, space-time image velocimetry, optical flow methods, and deep learning-based flow measurement methods in recent years. Finally, the existing challenges and issues were addressed, and potential future development directions were proposed.