From images to Modal Analysis via Phase-Based Motion Magnification and Estimation

In the aerospace field, very strict regulations require equally rigorous design and testing of structures and systems to ensure they meet high safety and performance standards. Compliance with these standards necessitates advanced testing and monitoring techniques. Experimental or Operational modal analysis serves this purpose and allows for the identification of the dynamic properties of structures, but it can also be employed for the validation and refinement of finite element models or structural health monitoring. A Ground Vibration Test is a perfect example of the implementation of EMA for the dynamic characterization of aerospace structures. In these tests, hundreds of accelerometers are placed over the structure to measure its response to the controlled excitation provided by the shakers. While these instruments are highly accurate, their drawbacks—such as the complexity of the setup, the mass-loading effect of attaching sensors to the structure, and the limitation of measurements being taken only at localized points—have led researchers to explore optical methods for non-contact, full-field measurements. Some examples include laser-based techniques such as Laser Doppler Vibrometry (LDV), as well as camera-based methods that rely mostly on Digital Image Correlation (DIC). ?? ??The present work explores a different image processing approach to the use of cameras as sensors. Motion enhancement techniques such as Phase-based Motion Magnification are used to isolate and magnify mode shapes to visualize them directly on the structure. The algorithm is based on Complex Steerable Decomposition to obtain the image local phase variation, proportional to the displacement of the structure. Once the phase information has been extracted, the frequency spectrum based on this data can be estimated to isolate single frequency bands associated with the mode shapes. The algorithm modifies the images to make the ODSs visible. The same information is then exploited to estimate the projection of the motion into the camera plane. When two cameras are used in a stereo configuration, the out-of-the-plane component of the motion is estimated using the triangulation technique. ?? ??In order to validate the algorithm, virtual images are created to replicate an experimental test. A FE model is used to calculate the response of a simple structure to an impact and the displacements are used to deform the mesh model in Blender, rendering the images of the deforming structure, using these images to validate the motion estimation model. Finally, experimental tests in a laboratory setting are conducted to evaluate the algorithm in real conditions. These experiments provide essential insights into the practical applicability and robustness of the method for motion enhancement and displacement measurement in experimental and operational modal analysis.