@Article{cmc.2026.086073,
AUTHOR = {You Yang, Bo Chen, Weiqi Liu, Zekai Ma},
TITLE = {Machine Vision-Based Super-Resolution Reconstruction for High-Precision Structural Displacement Monitoring: Potential Application to Hydraulic Structures},
JOURNAL = {Computers, Materials \& Continua},
VOLUME = {},
YEAR = {},
NUMBER = {},
PAGES = {{pages}},
URL = {http://www.techscience.com/cmc/online/detail/28240},
ISSN = {1546-2226},
ABSTRACT = {This paper proposes a non-contact intelligent method using machine vision and super-resolution (SR) reconstruction. The method uses artificial targets, a high-order degradation model, and camera-based acquisition for displacement calculation. Building upon the (Real-Enhanced Super-Resolution Generative Adversarial Networks) Real-ESRGAN framework, this paper introduces a feature fusion attention mechanism to improve the Real-ESRGAN network and generator, enabling the reconstruction of image contours and fine details to enhance displacement calculation accuracy. Quantitative laboratory validation on a benchmark shake-table dataset demonstrates improved displacement-monitoring accuracy, while hydraulic-structure imagery is used only to demonstrate reconstruction performance. The results indicate potential applicability to hydraulic-structure monitoring; independent quantitative validation under operational hydraulic conditions remains future work. Compared with traditional super-resolution algorithms and target tracking methods, the improved Real-ESRGAN algorithm performs the best, with a coefficient of determination (R<sup>2</sup>) of up to 0.9975, a mean absolute error (MAE) as low as 0.5552 mm, and residual errors controlled within 5 mm in the benchmark experiment. The reconstructed images preserve edge contours and fine details, supporting the use of super-resolution for marker-based structural displacement monitoring.},
DOI = {10.32604/cmc.2026.086073}
}