Displacement monitoring method of large structure based on dual-camera correction.

Visual measurement technology has the advantages of high precision, long-distance, multi-point, dynamic, and real-time measurement, and is widely used in the field of displacement monitoring of large structures. However, complex environmental conditions and construction disturbances can cause the vibration of measurement platform, resulting in significant measurement errors. To address this issue, we propose a displacement measurement method based on generalized dual camera relative motion estimation, and linear analytical solution of the method is derived. By forming a dual camera with a fixed connection angle, this approach improves the accuracy of translation and angle estimation along the optical axis compared to a single camera. It enables high-precision measurement of small vibrations in six degrees of freedom and compensates for the changes in the measurement point caused by platform motion. Simulation results show that our method outperforms monocular PnP algorithms, achieving accuracy comparable to binocular optimization methods while significant increasing the efficiency. Outdoor experiments demonstrate that it achieves a mean error of less than 1 mm compared to total station measurements, verifying its practicality and reliability.