Rapid vibration estimation in a single shot using white-light interferometry and deep learning

Environmental vibrations have a profound impact on the quality of interference fringe images and the overall accuracy of measurements in white-light interferometry. Due to the inherently high sensitivity of interferometric systems to optical path length variations at the nanometer scale, even the slightest mechanical vibrations can induce phase noise, distort, or shift the fringe image. These effects lead to reduced fringe contrast and clarity, making it difficult to accurately identify the zero-order fringe, a key factor in precise surface height measurement. As a consequence, measurement uncertainty increases, repeatability decreases, and systematic errors may arise in the reconstruction of three-dimensional surface profiles. We propose a rapid vibration estimation method using white-light interferometry based on a single fringe image and deep learning. The effectiveness of the proposed approach is validated through simulations and experiments. The results demonstrate the method’s strong potential for fast estimation of environmental vibration amplitudes and its applicability in providing auxiliary information for noise filtering algorithms.