Much information can be derived from operational deflection shapes of vibrating structures and the magnification of their motion. However, the acquisition of deflection shapes usually requires a manual definition of an object’s points of interest, while general motion magnification is computationally inefficient. We propose easy extraction of operational deflection shapes straight from vision data by analyzing and processing optical flow information from the video and then, based on these graphs, morphing source data to magnify the shape of deflection. We introduce several processing routines for automatic masking of the optical flow data and frame-wise information fusion. The method is tested based on data acquired both in numerical simulations and real-life experiments in which cantilever beams were subjected to excitation around their natural frequencies.
