| 基于背景差分的高铁钢轨表面缺陷图像分割 |
| |
| 引用本文： | 贺振东,王耀南,刘 洁,印 峰.基于背景差分的高铁钢轨表面缺陷图像分割[J].仪器仪表学报,2016,37(3):640-649. |
| |
| 作者姓名： | 贺振东 王耀南 刘 洁 印 峰 |
| |
| 作者单位： | 1.湖南大学电气与信息工程学院长沙410082；2. 郑州轻工业学院电气信息工程学院郑州450002,湖南大学电气与信息工程学院长沙410082,郑州轻工业学院电气信息工程学院郑州450002,湘潭大学信息工程学院湘潭411105 |
| |
| 基金项目： | 国家自然科学基金(60835004，6107212，6117216，61175075)、河南省科技攻关计划(42102210514,162102210060)项目资助 |
| |
| 摘 要： | 高铁钢轨表面图像具有光照变化、反射不均、特征少等特点,使得缺陷自动检测极为困难。为了在高速运动过程中,从复杂的钢轨表面图像中分割出缺陷,根据钢轨表面图像具有沿钢轨方向像素值基本不变的特征,建立钢轨表面图像背景模型,提出了基于背景差分的钢轨表面缺陷检测算法,主要包括钢轨区域提取、背景建模差分、阈值分割和图像滤波4个步骤,其主要特点是将视频监控中的背景差分法推广到缺陷图像分割领域,同时借助自适应阈值分割和滤波技术,在一定程度上,解决了铁轨表面缺陷分割过程中图像光照变化、反射不均、特征少等不利因素的影响。实验仿真和现场测试结果均表明,该方法对块状缺陷能很好地识别,召回率和准确率分别达96%和80.1%。
|
| 关 键 词： | 背景差分 自适应阈值 高铁钢轨表面缺陷 图像分割 |
Background differencing based high speed rail surface defect image segmentation |
| |
| He Zhendong,Wang Yaonan,Liu Jie and Yin Feng.Background differencing based high speed rail surface defect image segmentation[J].Chinese Journal of Scientific Instrument,2016,37(3):640-649. |
| |
| Authors: | He Zhendong Wang Yaonan Liu Jie and Yin Feng |
| |
| Abstract: | Illumination variation, reflection inequality and limited features existing in high speed rail surface images make the automated visual inspection task extremely difficult. In order to separate the defects from complex rail surface images in high speed motion process, according to the fact that the rail surface image has the characteristic of basically unchanged pixel value along the rail direction, the background model of the rail surface image is established. A rail surface defect detection algorithm based on background differencing is proposed. The algorithm contains four steps: extracting rail region, background modeling and differencing, threshold segmentation and image filtering. The main feature of the algorithm is extending the background differencing algorithm in video surveillance to defect image segmentation. By means of adaptive threshold segmentation and image filtering technology, the influences of image illumination variation, reflection inequality, limited features and other negative factors in the rail surface defect segmentation are decreased in certain degree. The simulation and field experiment results indicate that the proposed method can identify the block shape rail surface defects effectively. The recall and precision can reach 96% and 80.1%, respectively. |
| |
| Keywords: | |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《仪器仪表学报》浏览原始摘要信息 |
|
点击此处可从《仪器仪表学报》下载全文 |
|
