A two-stage mitigation method for optical turbulence-induced errors in vision-based structural displacement measurement

Abstract

Vision-based structural displacement measurement techniques have been widely applied. However, the visual sensors used for remote monitoring of structures in high-temperature weather are easily affected by optical turbulence, which introduces errors in displacement measurement. Therefore, this paper proposes a two-stage optical turbulence-induced error alleviation method. In the first stage, the steerable pyramid method is used to decompose the monitoring video and perform temporal filtering on the phase, which can significantly attenuate the motion and distortion caused by optical turbulence in the video. In the second stage, a feature point matching method considering the weighted distance is used to track the multi-point displacement in the reconstructed video to improve the robustness of feature point tracking, and the results are spatially filtered to improve measurement accuracy. The effectiveness of the proposed method has been verified through laboratory experiments and on-site testing.