Abstract In this paper, broad-band phase-based motion magnification (BPMM) is used to improve the modal parameter estimation from high-speed video of a structure undergoing low amplitude vibration. The authors use a novel application of the phase-based motion magnification technique to obtain improved results in the presence of image noise. A numerical simulation is performed to demonstrate and quantify the effect of broad-band motion magnification. A clear correlation is found between the magnification factor and the error in the motion obtained from 2-dimensional point tracking (2DPT). In a laboratory experiment, operational modal analysis is performed on a metallic nozzle with a 5.3 in exit diameter and a constant wall thickness of 0.03 inches. Fluorescent markers are painted on the nozzle lip. The nozzle is attached to a test stand and excited by pressurized air. The induced vibrations are captured by a single high-speed camera, which takes images of the nozzle lip. Using a 2DPT algorithm, the displacement history of the markers in the nozzle exit plane is extracted. As is typical for small and stiff structures, the resonant frequencies are quite high and the resulting vibration amplitudes are fairly low. This leads to low signal-to-noise ratio in the higher frequencies, which makes the following operational modal analysis much harder. By using BPMM as a preprocessor, the authors demonstrate that the outcome of modal analysis is greatly improved. Comparison with Finite Element analysis shows that the mode shapes agree much better with the experimental results when motion magnification is used. The modal parameters of the first six modes are found in the frequency range 0–1400 Hz using the new methodology, whereas without motion magnification only the first five modes are found.

中文翻译：

---

宽带相位运动放大对模态参数估计的影响

摘要 在本文中，宽带基于相位的运动放大（BPMM）被用来改进高速视频中低振幅振动结构的模态参数估计。作者使用基于相位的运动放大技术的新应用，在存在图像噪声的情况下获得改进的结果。进行数值模拟以证明和量化宽带运动放大的影响。在放大系数和从二维点跟踪 (2DPT) 获得的运动误差之间发现了明显的相关性。在实验室实验中，对出口直径为 5.3 英寸、壁厚恒定为 0.03 英寸的金属喷嘴进行了操作模态分析。荧光标记涂在喷嘴唇上。喷嘴安装在试验台上并由加压空气激发。诱发振动由单个高速摄像机捕获，该摄像机拍摄喷嘴唇部的图像。使用 2DPT 算法，提取喷嘴出口平面中标记的位移历史。对于小而硬的结构来说，共振频率非常高，产生的振动幅度非常低，这是典型的情况。这导致较高频率的低信噪比，这使得以下操作模态分析更加困难。通过使用 BPMM 作为预处理器，作者证明了模态分析的结果得到了极大的改善。与有限元分析的比较表明，当使用运动放大时，模态振型与实验结果更加吻合。