In recent years, video-based monitoring has become a well-established non-contact method for measuring river surface velocity and estimating cross-sectional flow. The basic principle involves calculating the pixel displacement velocity of water surface textures or particles, converting it into the actual velocity using the spatial scale corresponding to each pixel, and subsequently using this velocity as the water flow velocity. Accurate acquisition of pixel coordinates within the video frame is an essential process in video-based flow measurement, which directly affects the accuracy of the final velocity calculation. A comparison of several commonly used video calibration methods shows that current video calibration methods commonly suffer from low efficiency, insufficient accuracy, environmental constraints on target placement, and safety risks for personnel. Through theoretical analysis combined with practical application, the author proposes a laser positioning-based video calibration method tailored for scenarios without bank-based references or in complex river environments. This calibration method typically requires less than one hour per site, achieving centimeter-level accuracy. A case study of video calibration at the Xi Xiayuan Hydrological Station confirms that this method achieves an average error of 0.04 m over a hundred-meter distance and an average pixel error of 1.85 pixels, demonstrating a significant reduction in calibration error and improving the accuracy of video-based flow measurements. This paper provides a detailed discussion of this method from the perspectives of required equipment, calculation methodology, applicable environments, and key considerations.