基于图像处理的廊道表面裂缝检测技术研究  

Research on corridor surface crack detection technology based on image processing

作　　者：田伟 沈浩 李晓 师磊磊 TIAN Wei;SHEN Hao;LI Xiao;SHI Lei-lei(East China Tianhuangping Pumped Storage Co.,Ltd.,Huzhou 313302,China;Zhejiang University of Science and Technology,Hangzhou 310023,China)

机构地区：[1]华东天荒坪抽水蓄能有限责任公司,浙江湖州313302 [2]浙江科技学院,浙江杭州310023

出　　处：《电子设计工程》2020年第5期148-151,156,共5页Electronic Design Engineering

基　　金：国网新能源控股有限公司科研项目(SGXYTP00JHJS1800108)。

摘　　要：针对目前抽水蓄能电站廊道表面裂缝检测工作存在人工成本高、检测效率低等问题,文中研究了基于图像处理裂缝识别技术。该技术通过将RGB彩色图像转换为灰度图像,用线性变换方法对其进行增强处理,同时采用领域平均滤波法消除图像噪声的干扰。在此基础上,通过最佳阈值法对图像进行锐化处理,以突出裂缝特征。最后,利用裂缝图像边缘像素灰度沿法线方向阶跃变化的特征,以灰度值一阶导数为基础,采用Roberts边缘检测算子完成对裂缝特征的边缘检测。该研究为实现廊道表面裂缝的自动检测与高效识别提供理论支撑。Aiming at the problems of high labor cost and low detection efficiency in the detection of surface cracks in galleries of pumped storage power stations,this paper studies the technology of Crack Recognition Based on image processing.The technology transforms RGB color image into gray image,enhances it by linear transformation,and eliminates the interference of image noise by domain average filtering.On this basis,the image is sharpened by the best threshold method to highlight the crack characteristics.Finally,based on the first derivative of gray value,the edge detection operator Roberts is used to complete the edge detection of crack features by using the step change of gray level of edge pixels along the normal direction.This study provides theoretical support for the automatic detection and efficient identification of corridor surface cracks.

关 键 词：裂缝检测 图像处理 灰度图像 图像锐化 边缘检测 

分 类 号：TN919[电子电信—通信与信息系统] TM734[电子电信—信息与通信工程]