Structural vibration mode identification from high-speed camera footages using an adaptive spatial filtering approach

Single-Pixel Optical-Flow-Based Experimental Modal Analysis

“…13 it can be observed that the sharp edges of the pattern used (Fig. 12) are registered across two pixels, caused by the blurring effects of focus and shutter [28], leading to errors in the gradient intensities due to noise. If the gradient intensities |𝐼 ′ 0 | were calculated separately (20) by simply subtracting the low intensity value from the high intensity value divided by 2 px, e.g.…”

“…Theoretically, the total light intensity measured by a pixel is related to the number of photons captured by that pixel during the exposure time. It is important that the exposure time is as short as possible to avoid blurring effects [28]. A short exposure time can be achieved directly by exposing the object to strong light sources so that enough photons reach the sensor, or by increasing the gain (higher ISO numbers).…”

“…Yang and Dorn [26] (2020) obtained the structural response using the phase-based optical flow method [16] and then performed mode separation and mode-shape identification using the affinitypropagation-cluster method [27]. Li et al [28] (2022) developed an adaptive spatial filtering approach to the pixel intensities of sharp edges to fit the mode shapes and improve the modal displacements that are used for the identification of natural frequencies and damping ratios with the least-squares rational function method [29]. Wang et al [30] (2022) used the Lucas-Kanade method of optical flow [31] on randomly sampled video data to perform modal identification on modes lying above the Shannon-Nyquist sampling limit; modal parameters were identified during the mode reconstruction optimisation process.…”

Multi-level curvature-based parametrization and model updating using a 3D full-field response

“…Chen et al [22] used a full-field measurement to identify the dynamics of a non-linear structure. Li et al [23] used an adaptive filtering algorithm to enhance the signal power, decreasing the effect of the high noise level in the measurement. Javh et al [24] proposed a hybrid method, a combination of the simplified optical flow method with a high-dynamic-range sensor to identify the modal shapes under the noise floor of the Frequency Response Function (FRF).…”