Abstract: To address the challenges of measuring nuclear power pipeline vibrations due to mass loading effects and low spatial resolution, an optical flow algorithm considering data compensation for the pipeline vibration monitoring is proposed. This algorithm combines optical flow and camera calibration, and introduces data compensation for the issue of pipeline foundation vibration. Simulated pipeline vibration measurements are conducted in a laboratory setting, and the results demonstrate that the RMS's relative error of the algorithm proposed is less than 2.8% when compared with accelerometers and laser displacement sensors, and that the frequency's relative error is less than 2.2%. The robustness of the algorithm proposed is verified under ground noise conditions, and the necessity for data compensation is analyzed for the extreme scenarios involving flexible pipeline foundations. Moreover, lubricant pipeline vibration measurements and anomaly capture experiments are conducted in a nuclear power plant for on-site verification purpose. The results demonstrate that the RMS's relative error of the algorithm proposed is less than 4%, and that the frequency's relative error is less than 2.2%. The algorithm proposed can accurately measure the vibration response of the nuclear power pipeline under non-contact and marker-free conditions. It exhibits reliability in the presence of ground noise, and is capable of identifying vibration anomalies.