A two-stage correction method for UAV movement-induced errors in non-target computer vision-based displacement measurement

Abstract

Displacement plays a pivotal role in bridge assessment, but accurate displacement monitoring remains a challenging task. Unmanned Aerial Vehicles (UAVs) provide a cost-effective, time-efficient, and high maneuverability alternative to infrastructure monitoring, as they overcome the spatial limitations of the fixed camera and acquire the high-resolution image sequence. However, the measurement accuracy is often affected by the movement of the UAV. To address these constraints, this study proposed a computer vision-based nontarget displacement measurement method and a two-stage UAV movement correction method using fixed point and variational mode decomposition (VMD). Initially, the adaptive fusion of deep features and shallow features can efficiently encode the informative representation of the natural texture on the structural surface. Subsequently, the movement of the UAV is eliminated by stationary fixed points (Step Ⅰ) and VMD techniques (Step Ⅱ). Finally, the performance of the proposed methodology is verified with the field tests on a concrete wall and an arch bridge. Through mode decomposition and reconstruction, the measurement accuracy is greatly improved compared to the correction method only using fixed points, which proves the reliability and effectiveness of the proposed non-target displacement measurement method.