隧道检测裂缝的图像处理研究  

Study on Tunnel Crack Detection Based on Image Processing

作　　者：王睿[1] 漆泰岳[1] 朱鑫[1] 李涛[1] 

机构地区：[1]西南交通大学交通隧道工程教育部重点实验室,成都610031

出　　处：《铁道标准设计》2014年第10期93-96,127,共5页Railway Standard Design

基　　金：国家自然科学基金资助项目(51278423);中央高校基本科研业务费专项资金项目(SWJTU11ZT33)

摘　　要：裂缝是隧道衬砌最常见的病害之一,基于近几年快速发展的工程检测系统与图像处理算法的研究,提出基于CCD相机的衬砌裂缝检测系统来采集裂缝图像。通过比较均值滤波、中值滤波、维纳滤波、自适应中值滤波和加权邻域滤波比较,选择自适应中值滤波进行图像增强。结合直方图阈值分割法、Otsu最大类间方差阈值和局部阈值分割法,对增强后图像进行二值化处理比较,Otsu法较好地保留了裂缝的边缘信息,证明了该算法的有效性,为后续裂缝信息的提取奠定了基础。Cracks are one of the most common defects of the tunnel engineering. With rapid development of detection systems and image processing algorithms in recent years, a rapid lining inspection system based on CCD is proposed for image acquisition. By comparing the mean filter, median filter, Wiener filter, adaptive median filter and the algorithm of neighborhood weighted averaging, the author selects adaptive median filter for image enhancement. In combination with histogram threshold method, maximum between-cluster variance （Otsu） and local threshold segmentation to binarize and compare the enhanced images, the Otsu method better preserves the information about the edge of cracks, demonstrates the effectiveness of the algorithm, and lays a foundation for the extraction of further information about cracks.

关 键 词：隧道裂缝 裂缝图像 自适应中值滤波 Otsu最大类间方差 二值化 

分 类 号：U45[建筑科学—桥梁与隧道工程]