A Comparison Between Three Vision-Based Displacement Measurement Methods in Lab-Scale Structural Experiments

Abstract

Capturing structural displacement time-histories under service conditions or controlled load tests provides critical information for structural condition assessments. Measuring displacements directly is often impractical due to inherent technical and experimental difficulties, particularly for field inspection of in-service structures. Therefore, quantifying displacements indirectly from localized acceleration or strain measurements has been the dominant measurement approach for decades, despite numerical and logistical implementation challenges. More recently, the development of high-resolution imaging sensors, along with the emergence of high-performance digital image processing techniques, has led to the emergence of computer vision-based displacement tracking methods. These methods, generally referred to as optical flow field measurement, have been demonstrated to work effectively under a variety of laboratory and full-scale field conditions. The measurements are generally less accurate than those obtained from installed sensor arrays, but they have the advantage of providing dense displacement field measurements and do not require direct sensor installation. These approaches can be categorized into three major categories: dense flow, target or invariant feature tracking, and phase-based flow. This paper provides a comparative analysis of these three measurement paradigms for structural monitoring applications. A series of illustrative laboratory experiments are used to highlight key benefits and differences between different approaches.