基于MMSE的自适应灰度形态学钢轨边缘检测算法  

An adaptive gray morphological method for rail edge detection based on MMSE

作　　者：郭栋鸿 谭丽[1] 温润 GUO Dong-hong;TAN Li;WEN Run(College of Automation and Electrical Egineering,Lanzhou Jiaotong University,Lanzhou 730070,China;Project Management Center,State Grid Lanzhou Power Supply Company,Lanzhou 730070,China)

机构地区：[1]兰州交通大学自动化与电气工程学院,甘肃兰州730070 [2]国家电网兰州供电公司项目管理中心,甘肃兰州730070

出　　处：《云南大学学报（自然科学版）》2019年第6期1144-1151,共8页Journal of Yunnan University(Natural Sciences Edition)

基　　金：甘肃省教育厅自然科学基金(2017A-020);中国铁路总公司科技研究开发计划(2017J012-A)

摘　　要：钢轨边缘检测是铁路轨道异物入侵检测的关键技术,针对钢轨图像在采集过程中经常受到不同程度的噪声影响,以及传统边缘检测算法难以准确检测到钢轨边缘的问题,提出了一种基于MMSE (Multi-scale Multi-direction Structural Elements)的自适应灰度形态学钢轨边缘检测算法.首先根据轨道图像的噪声特点,使用多尺度结构元素的形态学滤波算法对轨道图像进行自适应滤波操作,实现钢轨边缘的增强和噪声的抑制;然后对滤波后的轨道图像使用多方向自适应灰度形态学边缘检测算子进行钢轨边缘检测.实验结果表明:该算法不仅有效滤除了采集图像中的噪声,而且能够较准确地检测到轨道图像中的钢轨边缘.Rail edge detection is the key technology of tracking foreign matter intrusion detection. In view of the rail images are often affected by different degrees of noise in the acquisition process and the traditional edge detection methods are difficult to accurately detect the rail edge, an adaptive gray morphological method for rail edge detection based on MMSE(multi-scale and multi-direction structural elements) is proposed in this paper.Firstly, according to the noise characteristics of the track images, the morphological filtering algorithm of multiscale structural elements are used to carry out the adaptive filtering operation on the track images, so as to enhance the rail edge and suppress the noise. Then the multi-direction adaptive gray morphological edge detection operator is used to detect the rail edge of the filtered track images. Experiments show that: This algorithm can not only effectively filter out the noise in the collected images, but also detect the rail edge in the track images accurately.

关 键 词：钢轨边缘检测 异物入侵 多尺度多方向 自适应滤波 灰度形态学 

分 类 号：U298.1[交通运输工程—交通运输规划与管理]