|
Compared with traditional measurement of structural vibration, which requires a limited number of sensors to be distributed discretely on the surface of the structure, vision-based vibration measurement makes use of each pixel as a potential measurement point. The increase number of sensors in spatial density by several orders of magnitude, thus achieving high spatial resolution. The machine vision measurement method based on motion amplification is used to measure the micro vibration of structures. The existing motion amplification algorithms rely on pyramid decomposition, which requires multi-scale and multi-direction analysis, and the noise reduction process is complicated. Therefore, to address these issues, this paper proposes a structural modal visualization algorithm based on two-step motion amplification. The visual measurement vibration method based on motion amplification can visualize the vibration mode of the structure under the condition of high resolution. Compared with other methods, we propose a simpler method than the vibration measurement extraction based on phase motion amplification and the blind signal separation of multi-directional steerable pyramids. This method uses singular value decomposition to achieve unified noise reduction of time-domain video in the pixel domain, and amplifies the decomposed signal through Euler amplification. At this time, there is still modal aliasing, and the mixed signal is decoupled. At this time, the separated signal is only related to the single mode of each order, and the signal is amplified based on the phase and the independent vibration mode is visualized. We verify our method through a micro-vibration video of a spherical geometric structure and a cantilever beam excitation experiment, and the results show that the method is effective.
|
