摘要
地铁隧道裂缝病害的自动检测技术是一个重要研究方向。针对地铁隧道复杂场景和弱光环境下,全局图像检测精度低的问题,提出分块图像局部纹理处理算法,将大视场裂缝图像进行网格化处理,在分块区域内完成预处理与纹理提取,基于图像细化与骨架提取算法,提出裂缝和虚假裂缝纹理的差异性计算模型,可有效提高真实裂缝图像的检测精度,滤除虚假裂缝的干扰。针对硬件系统,提出多目高速线阵相机的图像采集方案,研制裂缝图像采集系统样机,可安装于轨道小车上进行图像连续采集。利用研制的图像采集处理设备,可以自动采集和检测隧道裂缝图像,对于纹理简单的普通裂缝图像样本,裂缝的识别率达到0.96;对于地铁隧道裂缝图像样本,裂缝的识别率达到0.84,验证了硬件系统和软件算法的有效性与可行性。
Automatic detection of subway tunnel cracks is an important area of research.In response to the problem of low accuracy of global image detection in complex scene and weak light environment in subway tunnel,a local texture processing algorithm based on subdivided images was presented.By gridding of large view crack image,image preprocessing and texture extraction were completed in subdivided images.Based on the algorithm of image thinning and skeleton extraction,the difference calculation model of crack and false crack texture was proposed to increase the detection precision of the real crack image and eliminate the interference of false cracks.An image acquisition scheme based on a multiple linear array camera system was proposed and a prototype of the crack image acquisition system was developed,which can be mounted on a track vehicle for continuous image capturing.During the experiment,based on the developed image acquisition and processing device,automatic capturing and detection of tunnel crack images were implemented.For the simple texture of concrete crack images,the accuracy rate reaches 0.96.For the real crack images of subway tunnel,the accuracy rate is 0.84,verifying the effectiveness and feasibility of the hardware system and software algorithms.
出处
《铁道学报》
EI
CAS
CSCD
北大核心
2018年第2期82-90,共9页
Journal of the China Railway Society
基金
科技部国家重点研发计划(2016YFB1200402-002)
中央高校基本科研业务费(M16JB00240)
城市轨道交通系统安全与运维保障国家工程实验室项目
关键词
隧道裂缝
线阵相机
图像处理
裂缝检测
骨架提取
tunnel cracks
linear array camera
image processing
crack detection
skeleton extraction