基于数字图像处理的轨道板裂纹缺陷检测  

Research on the Algorithm for Crack Detection in Slab Track Based on Digital Image Processing

作　　者：吴成和[1] 卞春华[1] 朱挺[2] 马千里[3] 

机构地区：[1]南京大学电子科学与工程学院,江苏南京210093 [2]上海铁路局科学技术研究所,上海200071 [3]南京邮电大学地理与生物信息学院,图像处理与图像通信江苏省重点实验室,江苏南京210003

出　　处：《光电技术应用》2014年第1期49-51,64,共4页Electro-Optic Technology Application

基　　金：国家自然科学基金(60501003)

摘　　要：利用数字图像处理技术,通过轨道板图像自动分割、边缘提取、轨道板固有结构边缘去除、真伪裂纹区分等步骤,设计了轨道板裂纹缺陷自动检测识别算法。通过对实际采集的轨道板图像的处理,证明该方法能够准确提取和标记缺陷区域。在实验的1 000幅图像中,人工检出裂纹图像6幅,自动识别裂纹图像6幅,准确率达到100%,算法运行速度经测试能够满足实时检测的需求。The algorithm for automatic crack detection and identification in slab track images is designed based on digital image processing technology and many procedures such as image auto-segmentation, edge extrac-tion, edge removing of slab track structure and false crack identification. The algorithm has been proved accurately in extracting and marking the crack region by applying to slab track images from actual acquisition. In the experi-ment of 1000 images, all the six manually detected crack images are also detected automatically by the algorithm, i. e., the accuracy rate is 100%. The computational speed of the algorithm can satisfy the real-time detection require-ments.

关 键 词：图像处理 轨道板分割 裂纹检测 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]