摘要
针对人工检测混凝土建筑物裂缝危险性高、效率低、准确性低等问题,研究了一种能够检测混凝土建筑物早期裂缝的方法。运用计算机视觉技术采集混凝土裂缝的图像信息,通过卷积神经网络图像处理技术消除冗余噪声,基于Python-opencv图像处理方法对混凝土建筑物表面的裂缝进行提取与测量,最后通过Canny算子检验边缘与边界精确定位。结果表明:该方法准确性高,所得的裂缝数据精确度符合规范标准。
A method capable of detecting early cracks in concrete buildings is studied for the problems of high danger,low efficiency and low accuracy of manual detection of cracks in concrete structures.This study uses computer vision technology to collect image information of concrete cracks,eliminates redundant noise by convolutional neural network image processing technology,extracts and measures cracks on the surface of concrete structures based on Python-opencv image processing method,and finally precisely locates the cracks by checking the edges and boundaries with Canny operator.The results show that the method is highly accurate and the accuracy of the obtained crack data meets the specification standard.
作者
张慧慧
钟迪
Zhang Huihui;Zhong Di(School of Civil Engineering,Liaoning Technical University,Fuxin Liaoning 123000,China)
出处
《山西建筑》
2023年第13期183-187,共5页
Shanxi Architecture
基金
2022年辽宁省大学生创新创业训练计划项目:一种面向混凝土构筑物裂缝识别测量机器人的设计与实现(S202210147021)。
关键词
计算机视觉
裂缝检测
图像识别
设定阈值
卷积神经网络
computer vision
crack detection
image recognition
set threshold
convolutional neural network
