Computer vision-based real-time continuous monitoring of the pose for large-span bridge cable lifting structures

Abstract

Given the highly intricate terrain and environmental conditions in the mountainous region, the existing measurement methods fail to ensure high precision and real-time long-distance measurements. The present study proposes a methodology for real-time and continuous monitoring of the bridge cable lifting structure pose over long distances. The main contributions include: (1) Designing a framework for a monocular vision-based 6 degrees of freedom (6-DOF) pose monitoring system that enables real-time measurement of bridge cable lifting structure displacements and angles during the construction of large-span bridges. (2) Developing an improved approach to address potential mislocalization of coordinate origin in the sequence ordering method used in traditional corner detection when the target proportion relative to the captured image area is small. The effectiveness of the proposed method is verified through the theoretical simulation tests, laboratory experiments, and monitoring of the actual use of the cable lifting structure during the construction of the XiNing River Extra large bridge.