Comparing Feature Matching Algorithms for Measuring In-Plane Strains on Civil Infrastructures

Abstract

Maintaining structural integrity of civil infrastructures such as bridges and tunnels is always an essential task for civil engineers. Collapse or damage of these infrastructures may lead to a tremendous amount of painful injuries, casualties, and societal losses. This paper reported the work on evaluating optimal feature matching algorithms for development of visual sensing-based techniques to measure in-plane deflections and strains in order to facilitate monitoring and evaluation of integrity of civil infrastructures in a cost-effective way. A series of experiments were conducted in which three algorithms Digital Image Correlation (DIC), Scale-Invariant Feature Transform (SIFT) and Speeded-Up Robust Features (SURF) were compared. The result indicated that DIC has superiority among the three algorithms. To further assess the accuracy of DIC, a high-speed industrial camera was then used to capture a series of continuous image frames of deformed real-world scenarios. The DIC algorithm was adopted in the feature detection and tracking process, and in-plane displacement and strains were calculated and compared with the ground truth. The result indicated that the DIC-based method can achieve highly accurate performance in measuring in-plane deflections and strains for civil infrastructures and holds potential to the development of visual sensing enabled structural health monitoring.