高速铁路近景影像轨道边缘提取与匹配方法

发布时间：2018-04-30 04:30

本文选题：铁路轨道 + 边缘提取　；参考：《铁道学报》2017年08期

【摘要】：近景摄影测量应用于铁路轨道几何平顺性检测具有较大潜力,为获得轨道几何状态参数,需对轨道数字影像进行定向与建模,本文探索一种高速铁路近景影像轨道边缘提取方法并构建轨道边缘线的同名点坐标映射模型。根据轨道数字影像的灰度信息与点位分布特点,利用Canny算法和概率Hough变换提取铁路轨道的内边缘特征,采用ORB算法与KNN算法匹配轨道表面同名点,建立轨道表面同名点的坐标对应关系。利用边缘线与轨道表面几何关系,构建轨道边缘线的多项式映射模型以匹配出边缘同名点。试验结果表明,该方法在轨道边缘识别与同名点匹配方面有较高的可靠性,可为高速铁路轨道几何状态检测提供影像处理技术支持。
[Abstract]:Close range photogrammetry has great potential for railway track geometric smoothness detection. In order to obtain orbital geometric state parameters, it is necessary to orientate and model the orbital digital image. In this paper, a method of track edge extraction for high speed railway near view image is explored and the coordinate mapping model of the same point of track edge line is constructed. The features of gray information and point distribution of the image are used to extract the inner edge features of railway track by using Canny algorithm and probability Hough transform, and using the ORB algorithm and KNN algorithm to match the same point of the orbital surface to establish the coordinate correspondence of the same name points on the track surface. The mapping model is used to match the edges of the same name points. The experimental results show that the method has high reliability on the track edge recognition and the matching of the same name points. It can provide image processing technology for high-speed rail geometric state detection.

【作者单位】：西南交通大学测绘遥感信息系;
【基金】：国家自然科学基金(41472255)
【分类号】：TP391.41;U216.3

【相似文献】