基于计算机视觉的结构位移监测综述

梳理基于计算机视觉的结构位移监测方法的研究和应用进展，从系统组成、计算方法、影响因素和实际应用4个方面进行综述，探讨计算机视觉结构位移监测目前存在的研究不足并给出合理建议。在系统组成方面，介绍各种相机、镜头和标志物选择的标准及优缺点，给出相机和镜头的选择建议以及是否需要人工标志物的应用场合。在计算方法方面，介绍相机标定、特征提取、目标追踪和位移计算4个方面的实现方法，分析镜头畸变带来的测量误差和相机标定的必要性，介绍2种实用相机标定的简化方法。在影响因素方面，从硬件因素、图像处理算法和环境影响3个方面分析引起系统测量误差的来源，阐述减小系统误差的解决方案。在实际应用方面，介绍计算机视觉结构位移监测在结构状态评估中的应用，包括结构受力行为分析、承载力评估、模态参数识别、模型更新、损伤识别和索力估计。

Review of Computer Vision-based Structural Displacement Monitoring

Research and application advances of computer vision-based structural displacement monitoring methods were addressed in this paper. Furthermore, overviews of four aspects, system configuration, computing methods, influence factors, and real applications, were provided. The current research limitations of computer vision-based structural displacement monitoring were explored and reasonable suggestions were provided. With respect to system configuration, the selection criteria and merits and demerits of various cameras, lenses, and markers were presented. Suggestions regarding selection of cameras and lenses were provided, and different scenarios of manual markers were presented. With regard to computing methods, the realization methods of camera calibration, feature extraction, target tracking, and displacement calculation were introduced. Additionally, the measurement error induced by lens distortion and the necessity of camera calibration were analyzed, and two practical and simplified camera calibration methods were introduced. Regarding influence factors, the sources of systematic measurement errors caused by hardware, image processing algorithms, and environmental factors were analyzed, and solutions for mitigation of systematic errors were elaborated. With regard to real applications, the applications of computer vision-based structural displacement monitoring in structural condition assessment were introduced, which included structural behavior analysis, loading capacity evaluation, modal parameter identification, model updating, damage detection, and cable force estimation.