摘要
对钢轨表面缺陷检测方法进行了研究,报告了检测方法的现状。提出了基于图像间差异的方法,快速从海量的钢轨表面图像中找到拟缺陷图像,过滤掉正常的图像。提出了基于视觉反差的显著图生成方法,准确提取缺陷的大小,位置等信息。实验结果表明,基于图像差异和视觉反差的钢轨表面缺陷检测方法具有速度快,漏检率低,误检率低的优点,能满足高速检测的需求。
The rail surface defects has important influence on the safety of train operation.The rail surface defect detection method was studied and the detection method of the status was reported.A method based on the difference between the images was proposed to quickly find the rail surface to be defected from the mass images and to filter the defect-free image.A method for generating saliency map based on visual contrast was proposed to accurate extraction of defect size,location and other information.The experimental results showed that,the surface defect detection method based on difference image and visual contrast with fast speed,low false negative rate,and low false positive rate,could meet the demand of high speed detection of rail surface.
出处
《计算机工程与设计》
CSCD
北大核心
2014年第6期2052-2055,共4页
Computer Engineering and Design
基金
国家自然科学基金项目(50808025)
湖南省科技计划基金项目(2013GK3012)
关键词
钢轨表面
图像间差异
拟缺陷图像
视觉反差
显著图
rail surface
image difference
possibility defect image
visual contrast
saliency map
