基于机器视觉的钢轨表面损伤检测研究  

Research on rail surface damage detection based on machine vision

作　　者：曾树华[1] 黄银秀[2] ZENG Shu-hua;HUANG Yin-xiu(Hunan Vocational College of Railway Technology,Zhuzhou 412006,China;HuNan Chemical Vocational Technology College,Zhuzhou Hunan 412006)

机构地区：[1]湖南铁路科技职业技术学院,湖南株洲412006 [2]湖南化工职业技术学院,湖南株洲412001

出　　处：《电子产品世界》2021年第12期53-54,共2页Electronic Engineering & Product World

基　　金：湖南省教育厅资助科研项目(19C1214)。

摘　　要：目的:为探寻钢轨表面伤损机器视觉检测有效方法,验证经典的5种边缘检测算法在钢轨表面损伤检测;方法:利用采集的带扎伤钢轨图片,在MATLAB2016版本进行检测实验;结果:log检测法、Canny检测法存在错检,roberts检测法、prewitt检测法、sobe检测法错在漏检;结论:经典的5种边缘检测算法在钢轨表面损伤检测无法达到工业级效果。Objective:In order to explore an effective method of machine vision detection of rail surface damage,five classical edge detection algorithms are verified in rail surface damage detection;Methods:Using the collected rail pictures with puncture injury,the detection experiment was carried out in MATLAB2016 version;Results:Log test and canny test were wrong,but Roberts test,Prewitt test and SOBE test were wrong;Conclusion:The five classical edge detection algorithms can not achieve industrial effect in rail surface damage detection.

关 键 词：机器视觉 边缘检测算法 钢轨表面损伤 

分 类 号：U216.3[交通运输工程—道路与铁道工程] TP391.41[自动化与计算机技术—计算机应用技术]