Structural health monitoring techniques have been applied to several important structures and infrastructure facilities, such as buildings, bridges, and power plants. For buildings, accelerometers are commonly used for monitoring the accelerations induced by ambient vibration to analyze the structural natural frequencies for further system identification and damage detection. However, due to the relatively high cost of the accelerometers and data acquisition systems, accelerometer-based structural health monitoring systems are challenging to deploy in general buildings. This study proposed an image analysis-based building deformation monitoring method that integrates a small single-board computer, computer vision techniques, and a single-camera multiple degree-of-freedom algorithm. In contrast to other vision-based systems that use multiple expensive cameras, this method is designed for a single camera configuration to simplify the installation and maintenance procedures for practical applications. It is designed to monitor the inter-story drifts and torsional responses between the ceiling and floor of a story that is being monitored in a building, aiming to maximize the monitored structural responses. A series of 1:10 reduced scale static and dynamic structural experiments demonstrated that the proposed method and the device prototype are capable of analyzing images and structural responses with an accuracy of 0.07 and 0.3 mm from the results of the static and dynamic experiments, respectively. As digital imaging technology has been developing dramatically, the accuracy and the sampling rates of this method can be improved accordingly with the development of the required hardware, making this method practically feasible for an increasing number of applications for building structural monitoring.

中文翻译：

---

图像分析在建筑物间漂移监测中的应用

结构健康监测技术已应用于一些重要的结构和基础设施，例如建筑物，桥梁和发电厂。对于建筑物，加速度计通常用于监视环境振动引起的加速度，以分析结构固有频率，以进一步进行系统识别和损坏检测。但是，由于加速度计和数据采集系统的成本较高，因此基于加速度计的结构健康监测系统很难在一般建筑物中部署。这项研究提出了一种基于图像分析的建筑变形监测方法，该方法集成了小型单板计算机，计算机视觉技术和单摄像机多自由度算法。与使用多个昂贵摄像头的其他基于视觉的系统相比，此方法设计用于单个摄像头配置，以简化实际应用中的安装和维护过程。它旨在监视建筑物中正在监视的楼层的天花板和地板之间的层间偏移和扭转响应，旨在最大程度地监视受监视的结构响应。一系列的1:10缩小比例的静态和动态结构实验表明，所提出的方法和设备原型能够从静态和动态实验的结果中分别以0.07和0.3 mm的精度分析图像和结构响应。随着数字成像技术的飞速发展，