基于计算机视觉的往复荷载作用下钢筋混凝土柱裂缝监测

Crack Monitoring of RC Columns Under Cyclic Loading Based on Computer Vision

通过对图像拼接得到大范围裂缝图像,采用阈值分割算法进行裂缝识别,从而建立起基于计算机视觉的混凝土构件表面大范围裂缝识别流程。在此基础上,对往复荷载作用下的钢筋混凝土柱进行裂缝监测,研究加载过程中混凝土柱表面裂缝开展情况。结果表明:采用图像拼接,可以利用消费级手机分区获取裂缝图像,进而识别混凝土构件表面大范围的裂缝;所建立的裂缝识别流程能较好地识别混凝土柱上大范围表面裂缝,裂缝宽度、长度和倾角识别精度较高;随着位移幅值的增加,混凝土柱表面裂缝长度、宽度和面积增加,倾角减小,损伤随之增加。

Based on computer vision,the procedure of recognizing wide-range cracks on the surface of concrete members was presented by obtaining the wide-range crack images through image stitching,and identifying cracks by the threshold segmentation.Adopting the procedure,crack monitoring was conducted on a reinforced concrete(RC)column under cyclic loading to study crack development on the surface during loading and unloading processes.The results showed that wide-range concrete surfaces with cracks could be photographed in a sub-regional way with consumer-grade mobile phones.By further stitching the photos together,the cracks were finely recognized.The proposed procedure could be well used to recognize wide-range cracks on the surface of RC columns with relatively higher identification accuracy regarding crack width,length and inclined angles.In addition,with the increase of displacement amplifude,the width,length and areas of cracks on the surface of the RC column increased,the inclined angles decreased,and then the damage increased.