A Visual Measurement Method for Large-Span Bridge Deformation Combining Super-Resolution-Aided and Rotating Box Target Tracking

Although vision-based measurement methods are beginning to be widely used in measuring displacements of structural bodies, limitations such as low camera resolution or difficulty in selecting the location of the measurement base station still exist in vision-based structural displacement measurement methods. To address the above problems, this paper proposes a combined super-resolution-assisted and rotating box target tracking visual measurement method for the deformation of large-span bridges. Through several effective innovations, the accuracy of the visual algorithm on the displacement measurement of the structure body is greatly improved based on solving the deficiencies of the existing visual algorithm. The algorithm in this paper reduces the mRMSE by 7.64%, the mMAE by 12.72%, and the mR<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> improves by 0.69% compared with the baseline YOLOv9 network. We conducted accurate experimental tests on a model bridge and then applied this method to the Longjiang Bridge and Humen Bridge, and all three evaluation indexes were optimal. Our code will be available at https://github.com/Onlyou-wu/SRTNet-YOLO_SR.