Abstract
Aiming at the problem that low-quality images and low-precision control points lead to scale differences between the survey area model and the real model in UAV (Unmanned Aerial Vehicle) vision-based 3D deformation monitoring, which impairs the accuracy of deformation monitoring, this paper develops a spatial 3D scale for providing high-precision scale information and proposes a UAV vision-based deformation monitoring method with 3D scale constraints, thereby improving the deformation monitoring accuracy in large-scale survey areas. Experimental results show that compared with the monitoring method using only control points as constraints, the proposed method achieves accuracy (RMSE) improvement rates of 38.6% and 48.1% in the horizontal and elevation directions respectively during four phases of UAV operations, and the 3D deformation accuracy (RMSE) improvement rate remains at approximately 42.3% during seven phases of UAV operations. This verifies the effectiveness and reliability of the UAV vision-based deformation monitoring method with 3D scale constraints.