Non-contact vibration analysis using speckle-based techniques

Abstract

In this work, two different speckle based techniques are employed to quantify the movement of a vibrating surface. Directly projecting a coherent laser to a rough surface and collecting its associated subjective speckle using a high-speed camera, different frequencies are detected successfully. For the same purpose, a speckle pattern created by launching a coherent laser within a multimode fiber is employed as structured light to detect changes into the projection surface. The achieved results suggest that subjective speckle is better to analyze small sized areas, even at relative high frequencies, while projected speckle can give information about wider areas.

Remote recovery of audio signals from videos of optical speckle patterns: a comparative study of signal recovery algorithms

“…We have compared six signal recovery algorithms in terms of the quality of the recovered signal and the computational costs. We have considered the differential processing method used in [11,12] which is simple to implement, but introduces undesired nonlinearities. We have analyzed how the mean intensity of the speckle pattern can recover the audio signal and have shown that a single observed pixel, if it is appropriately selected (the pixel with largest variance) can be sufficient for obtaining a reasonably good recovery of the audio signal.…”

“…Diode lasers are particularly attractive for sensing applications due to the fact that they are compact, energy-efficient, low-cost, and cover a wide range of wavelengths. Because of these advantages, here we implement the setup proposed by Rodriguez-Cobo et al [11] and Robles-Urquijo et al [12] for the remote recovery of audio signals from videos of speckle patterns [13], but using a diode laser. In our setup the output of the laser is projected to the membrane of a loudspeaker and the movement of the speckle pattern when the speaker plays a sound is video-recorded with a CCD camera.…”

“…Our goal is to perform a comparative study of different signal processing algorithms for the recovery of the audio signal from the silent video. We propose new recovery algorithms and compare with those that have been proposed in the literature [12,14]. The technique that was proposed by researchers at MIT [14] allows to extract sound from videos of minute vibrations.…”

“…The technique that was proposed by researchers at MIT [14] allows to extract sound from videos of minute vibrations. It does not use laser-generated speckle, and therefore, the video scene needs to be naturally illuminated (in contrast, here we illuminate the video scene with a laser from a distance, as in [12]). Another technique proposed here uses the algorithm IsoMap [15] for dimensionality reduction, that reconstructs time-varying features in a low-dimensional space.…”