钢轨轮廓全断面检测中的快速高鲁棒性匹配方法研究  

Research on Fast and Robust Matching Algorithm in Inspection of Full Cross-section Rail Profile

作　　者：冯凯 于龙[1] 占栋[1] 张冬凯 FENG Kai;YU Long;ZHAN Dong;ZHANG Dongkai(School of Electrical Engineering,Southwest Jiaotong University,Chengdu 610031,China)

机构地区：[1]西南交通大学电气工程学院

出　　处：《铁道学报》2019年第5期162-167,共6页Journal of the China Railway Society

基　　金：国家自然科学基金(U1234203)

摘　　要：采用激光摄像技术对钢轨全断面廓形进行检测,为保证检测数据的准确性和实时性,其关键在于钢轨廓形的快速高鲁棒性匹配算法。在分析国内外钢轨轮廓检测、匹配现状的基础上,对钢轨廓形匹配方法进行了系统研究。根据标准钢轨不同半径滚动圆空间几何分布特性,提出利用钢轨廓形的斜率切线值来对钢轨原始廓形轨腰曲线部分进行自动分段,结合最小二乘拟合算法,处理分段后的钢轨廓形,快速完成钢轨廓形初匹配。通过改进ICP算法,完成钢轨廓形二维点云的精确匹配,缩减了匹配时间,提高了匹配鲁棒性。最后,将该方法应用于轨道检测设备的数据采集中,验证了该方法的有效性。During the use of the line structured light vision sensor to detect the full cross-section rail profile shape,in order to ensure the accuracy and real-time performance of the measured data,the key lies in the fast and robust matching algorithm for rail profile. Based on the analysis of current domestic and international situation of rail profile detection,an rail profile matching method was put forward. According to the spatial geometric characteristics of different rolling circle slope on the standard rail profile,the automatic segmentation method was carried out where the slope tangent value of rail profile was used to automatically segment the waist curve of the original rail profile. The least-square fitting algorithm was used to pro-cess the segmented rail profile,to quickly complete the initial matching of rail profile. By improving the ICP (Iterative Close Point) algorithm,accurate matching of two-dimensional point cloud of rail profile is achieved,with shortened matching time and improved matching stability. Finally,the proposed approach proved to be practical after it was applied to the data acquisition of rail profile detection equipment.

关 键 词：全断面 高精度 快速 高鲁棒性 动态测量 

分 类 号：U216.3[交通运输工程—道路与铁道工程]