Robust Discriminative Correlation-Based Full-Field Motion Estimation of Large-Scale Structures Using One Mono Camera

Abstract

Full-field motion, reflecting the health state of large-scale structures, is hardly obtained by traditional structural health monitoring (SHM) systems due to limited measurement points. Besides, many structures are not equipped with an SHM system monitoring accurate motions. The easily-accessible surveillance videos with numerous pixels working as a bunch of sensors are thus promising in estimating full-field motion with a high resolution. However, existing vision-based motion estimation methods fail to achieve good fidelity and robustness, simultaneously. Therefore, a novel motion estimation method is proposed to measure the full-field motion of large-scale structures using one mono camera. In this approach, robust discriminative correlation is adopted to detect targets with various shapes by adaptively filtering the textures of the target and the background. The fidelity of the estimated motion reaches subpixel by introducing continuous convolution, which transforms the discrete pixel into a continuous function. A factorized convolution operator and a Gaussian mixture model are used to compact the number of model parameters and training samples, respectively. The present approach estimates the accurate displacement and high-resolution mode shape in an experimental study. Moreover, the displacement of the antenna on the high-rise Saige Building is estimated using a portal camera, which matches well with that using a laser Doppler vibrometer and further visualizes the high-resolution mode shapes of the antenna. The full-field vortex-excited resonance of the long-span Humen Bridge is estimated using the surveillance video, yielding accurate mode shapes agreeing well with the accelerometer results.

Keywords: Full-field motion, discriminative correlation, large-scale structure, mono camera