A study on real-time monitoring method based on digital twin and video fusion is carried on to address key technical challenges in real-time monitoring of surface flow pattern in channels,such as difficulty in fusing multi-source information,insufficient accuracy in correcting geometric distortions in videos,and inaccurate dynamic mapping of 3D scenes.By developing panoramic real-time video surveillance coupling technology,the compatibility problem of heterogeneous device data is solved,and the video stream access delay is reduced to less than 0.5 s.By developing an intelligent fusion algorithm for surface flow monitoring videos,the pixel level matching error between monitoring images and 3D scenes is achieved with less than 0.1 pixels.On-site testing of the surface flow state intelligent perception system is carried out in the Three Gorges channel of the Yangtze River,achieving an automatic recognition accuracy of 95.3% for typical flow states such as rapids and backflow,and controlling the flow velocity measurement error within ±0.05 m/s.The results indicate that the proposed intelligent fusion algorithm and system architecture significantly improve the real-time and accuracy of channel flow monitoring,providing reliable technical support for navigation safety management under complex hydrological conditions.