Date of Award


Publication Type

Master Thesis

Degree Name



Civil and Environmental Engineering


cable-stayed bridges, digital image processing, feature-based detection, non-contact vision-based measurement techniques, stay cables, tensile forces


Shaohong Cheng


Faouzi Ghrib




In cable-stayed bridges, the structural systems of tensioned cables play a critical role in structural and functional integrity. Thereby, tensile forces in the cables become one of the essential indicators in structural health monitoring (SHM). In this thesis, a video image processing technology integrated with cable dynamic analysis is proposed as a non-contact vision-based measurement technique, which provides a user-friendly, cost-effective, and computationally efficient solution to displacement extraction, frequency identification, and cable tension monitoring. In contrast to conventional contact sensors, the vision-based system is capable of taking remote measurements of cable dynamic response while having flexible sensing capability. Since cable detection is a substantial step in displacement extraction, a comprehensive study on the feasibility of the adopted feature detector is conducted under various testing scenarios. The performance of the feature detector is quantified by developing evaluation parameters. Enhancement methods for the feature detector in cable detection are investigated as well under complex testing environments. Threshold-dependent image matching approaches, which optimize the functionality of the feature-based video image processing technology, is proposed for noise-free and noisy background scenarios. The vision-based system is validated through experimental studies of free vibration tests on a single undamped cable in laboratory settings. The maximum percentage difference of the identified cable fundamental frequency is found to be 0.74% compared with accelerometer readings, while the maximum percentage difference of the estimated cable tensile force is 4.64% compared to direct measurement by a load cell.