Structural health monitoring at local and global levels using computer vision technologies has gained much attention in the structural health monitoring community in research and practice. Due to the computer vision technology application advantages such as non-contact, long distance, rapid, low cost and labor, and low interference to the daily operation of structures, it is promising to consider computer vision–structural health monitoring as a complement to the conventional structural health monitoring. This article presents a general overview of the concepts, approaches, and real-life practice of computer vision–structural health monitoring along with some relevant literature that is rapidly accumulating. The computer vision–structural health monitoring covered in this article at local level includes applications such as crack, spalling, delamination, rust, and loose bolt detection. At the global level, applications include displacement measurement, structural behavior analysis, vibration serviceability, modal identification, model updating, damage detection, cable force monitoring, load factor estimation, and structural identification using input–output information. The current research studies and applications of computer vision–structural health monitoring mainly focus on the implementation and integration of two-dimensional computer vision techniques to solve structural health monitoring problems and the projective geometry methods implemented are utilized to convert the three-dimensional problems into two-dimensional problems. This review mainly puts emphasis on two-dimensional computer vision–structural health monitoring applications. Subsequently, a brief review of representative developments of three-dimensional computer vision in the area of civil engineering is presented along with the challenges and opportunities of two-dimensional and three-dimensional computer vision–structural health monitoring. Finally, the article presents a forward look to the future of computer vision–structural health monitoring.

A newly developed, completely contactless structural health monitoring system framework based on the use of regular cameras and computer vision techniques is introduced for obtaining displacements and vibrations of structures, which are critical responses for performance-based design and evaluation of structures. To provide contactless and practical monitoring, the current vision-based displacement measurement methods are improved by eliminating the physical target attachment. This is achieved by means of utilizing imaging key-points as virtual targets. As a result, pixel-based displacements of a monitored structural location are determined by using an improved detection and match key-points algorithm, in which false matches are identified and discarded almost completely. To transform pixel-based displacements to engineering units, a practical camera calibration method is developed because calibration standard on a physical target no longer exists. Moreover, a framework for evaluating the accuracy of vision-based displacement measurements is established for the first time, which, in return, provides users with the most crucial information of a measurement. The proposed framework along with a conventional sensor network and a data acquisition system are applied and verified on a real-life stadium during football games for structural assessment. The results obtained by the new method are successfully validated with the data acquired from sensors such as linear variable differential transformers and accelerometers. Because the proposed method does not require any type of sensor and target attachment, common field works such as sensor installation, wiring, maintaining conventional data acquisition systems are not required. This advantage enables an inexpensive and practical way for structural assessment, especially for real-life structures. Copyright © 2016 John Wiley & Sons, Ltd.

Current vision-based displacement measurement methods have limitations such as being in need of manual targets and parameter adjustment, and significant user involvement to reach the desired result. This study proposes a novel structural displacement measurement method using deep learning-based full field optical flow methods. The performance of the proposed method is verified via a laboratory experiment conducted on a grandstand structure with a comparative study, where the same data samples are analysed with a commonly used vision-based method, and a displacement sensor measurement is used as the ground truth. Statistical analysis of the comparative results show that the proposed method gives higher accuracy than the traditional optical flow algorithm and shows consistent results in compliance with displacement sensor measurements. Image collection, tracking, and non-uniform sampling are investigated in the experimental data and suggestions are made to obtain more accurate displacement measurements. A field-validation on a footbridge showed that the measurement error induced by the camera motion is mitigated by a camera motion subtraction procedure. The proposed method has good potential to be applied by structural engineers, who have little or no experience in computer vision and image processing, to do vision-based displacement measurements.

Critical issues related to damage and condition assessment are discussed. The distinction between global and local state properties, linearized and nonlinear condition indices and experimental constraints are discussed along with a classification of condition assessment techniques. Brief examples from research results on steel stringer highway bridges and a long span bridge are presented in order to exemplify the interrelation between structural condition, damage, and some of the promising damage indices. The indices that are discussed in this paper are influence coefficients for displacement and strain, and redistribution of intrinsic strains measured by continuous long term monitoring. A discussion of these indices is introduced with application examples from the bridges, which were utilized as a test bed to evaluate the applicability of the indices. Tools and technologies for successful measurement of influence coefficients are presented. It is shown that if a spectrum of appropriate experiments and in...