Abstract This paper develops vision-based measurement methods for experimental tests of reinforced concrete (RC) structures. The methods can measure deformations and characterize cracks from images of RC specimens. The coordinates of objects of interest (OOIs) in the specimen are measured using a target tracking approach, and then deformation components (e.g., flexural, shear and sliding deformations) of the specimen are computed from the coordinates of OOIs through geometry analysis. The cracks are (i) identified using binary images converted from color images, (ii) and then quantified using the filter-based algorithm. The morphological operations, separation algorithm and connected component labeling algorithm are used in the image processing for crack measurements. The developed vision-based measurement methods were applied to cyclic tests of RC wall specimens. The accuracy of the vision-based measurements was validated by comparison with the results of traditional measurement techniques using the displacement transducers and crack scales. The proposed vision-based measurement methods demonstrate much higher efficiency and provide more useful information than the traditional measurement techniques. The paper also discusses a few application issues such as the specimen surface requirements and resolution of the vision-based measurements.

中文翻译：

---

用于 RC 结构测试的基于视觉的变形和裂纹测量

摘要 本文开发了基于视觉的测量方法，用于钢筋混凝土 (RC) 结构的实验测试。该方法可以测量变形并从 RC 试样的图像中表征裂纹。使用目标跟踪方法测量样本中感兴趣对象 (OOI) 的坐标，然后通过几何分析从 OOI 的坐标计算样本的变形分量（例如，弯曲、剪切和滑动变形）。裂缝是 (i) 使用从彩色图像转换而来的二值图像识别，(ii) 然后使用基于过滤器的算法进行量化。裂纹测量的图像处理采用形态学操作、分离算法和连通分量标记算法。开发的基于视觉的测量方法被应用于钢筋混凝土墙试样的循环测试。通过与使用位移传感器和裂纹尺度的传统测量技术的结果进行比较，验证了基于视觉的测量的准确性。所提出的基于视觉的测量方法比传统的测量技术具有更高的效率并提供更多有用的信息。本文还讨论了一些应用问题，例如试样表面要求和基于视觉的测量的分辨率。所提出的基于视觉的测量方法比传统的测量技术具有更高的效率并提供更多有用的信息。本文还讨论了一些应用问题，例如试样表面要求和基于视觉的测量的分辨率。所提出的基于视觉的测量方法比传统的测量技术具有更高的效率并提供更多有用的信息。本文还讨论了一些应用问题，例如试样表面要求和基于视觉的测量的分辨率。