基于背景差分的高铁钢轨表面缺陷图像分割
发布时间：2018-11-07 09:50
【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%。
[Abstract]:The surface image of high-speed rail is characterized by light variation, uneven reflection and less characteristics, which makes automatic defect detection very difficult. In order to segment defects from complex rail surface images during high-speed motion, a background model of rail surface images is established according to the fact that the rail surface images have the same pixel value along the rail direction. A rail surface defect detection algorithm based on background difference is proposed, which includes four steps: rail region extraction, background modeling difference, threshold segmentation and image filtering. The main feature of this method is that the background differential method in video surveillance is extended to the field of defect image segmentation. At the same time, with the help of adaptive threshold segmentation and filtering technology, the illumination variation of the image in the course of rail surface defect segmentation is solved to a certain extent. Uneven reflection, less characteristics and other adverse factors. The experimental results and field test results show that the proposed method can recognize the block defects well, and the recall rate and accuracy are 96% and 80.1%, respectively.
【作者单位】： 湖南大学电气与信息工程学院;郑州轻工业学院电气信息工程学院;湘潭大学信息工程学院;
【基金】：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
【分类号】：U216.3;TP391.41
，
本文编号：2315980
[Abstract]:The surface image of high-speed rail is characterized by light variation, uneven reflection and less characteristics, which makes automatic defect detection very difficult. In order to segment defects from complex rail surface images during high-speed motion, a background model of rail surface images is established according to the fact that the rail surface images have the same pixel value along the rail direction. A rail surface defect detection algorithm based on background difference is proposed, which includes four steps: rail region extraction, background modeling difference, threshold segmentation and image filtering. The main feature of this method is that the background differential method in video surveillance is extended to the field of defect image segmentation. At the same time, with the help of adaptive threshold segmentation and filtering technology, the illumination variation of the image in the course of rail surface defect segmentation is solved to a certain extent. Uneven reflection, less characteristics and other adverse factors. The experimental results and field test results show that the proposed method can recognize the block defects well, and the recall rate and accuracy are 96% and 80.1%, respectively.
【作者单位】： 湖南大学电气与信息工程学院;郑州轻工业学院电气信息工程学院;湘潭大学信息工程学院;
【基金】：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
【分类号】：U216.3;TP391.41
，
本文编号：2315980
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2315980.html
