Phase-based motion processing and the associated Motion Magnification that it enables has become popular not only for the striking videos that it can produce of traditionally stiff structures visualized with very large deflections, but also for its ability to pull information out of the noise floor of images so that they can be processed with more traditional optical techniques such as digital image correlation or feature tracking. While the majority of papers in the literature have utilized the Phase-based Image Processing approach as a pre-processor for more quantitative analyses, the technique itself can be used directly to extract modal parameters from an image, noting that the extracted phases are proportional to displacements in the image. Once phases are extracted, they can be fit using traditional experimental modal analysis techniques. This produces a mode “shape” where the degrees of freedom are phases instead of physical motions. These phases can be scaled to produce on-image visualizations of the mode shapes, rather than operational shapes produced by bandpass filtering. Modal filtering techniques can also be used to visualize motions from an environment on an image using the modal phases as a basis for the expansion.

中文翻译：

---

使用来自基于相位的运动处理和运动放大的相位量的实验模态分析

基于相位的运动处理和它启用的相关运动放大已经变得流行，不仅因为它可以制作具有非常大偏转可视化的传统刚性结构的引人注目的视频，而且还因为它能够从噪声本底中提取信息图像，以便它们可以使用更传统的光学技术进行处理，例如数字图像相关或特征跟踪。虽然文献中的大多数论文都利用基于相位的图像处理方法作为更多定量分析的预处理器，但该技术本身可以直接用于从图像中提取模态参数，注意提取的相位与图像中的位移。一旦提取了相，它们就可以使用传统的实验模态分析技术进行拟合。这会产生一种模式“形状”，其中自由度是相位而不是物理运动。这些相位可以缩放以产生模式形状的图像可视化，而不是带通滤波产生的操作形状。模态滤波技术也可用于使用模态相位作为扩展的基础来可视化来自图像上的环境的运动。