Stereo-point tracking of inherent structural features for 3D computer vision measurements

Computer vision techniques have gained popularity for structural health monitoring and structural dynamics thanks to advancements in camera technology and computational capability. Among these techniques, three-dimensional-digital image correlation (3D-DIC) and 3D-point tracking (3D-PT) have been used as substitutes for traditional contact-based methods. To produce accurate measurements, 3D-DIC and 3D-PT require high-contrast patterns such as stochastic speckles or fabricated optical targets on the surface of the targeted structure. However, when large-scale engineering structures such as bridges and wind turbines must be analyzed, applying the high-contrast speckles/targets may not always be feasible, thus limiting the applicability of 3D-DIC and 3D-PT. This research develops an approach to identify and track inherent features such as bolts, letters, stains, rusted patches, and holes that are already present on the structure. To achieve this goal, the Augmented Convex Polygon of Gradients (ACPG), an algorithm based on a novel custom interpolation method, is developed to track the motion of inherent structural features. In this research, the performance of the proposed ACPG algorithm is compared with traditional 3D-PT measurements and contact-based sensors in laboratory and field tests. Results show that measurements obtained with the ACPG algorithm are in excellent agreement with those obtained with traditional approaches, yielding accuracy between 93 % and 99 %. The outcomes of this research prove that the proposed ACPG algorithm can extend the applicability of computer vision techniques for 3D measurements of real-world structures.