铁路轨道近景影像特征的自动识别与无缝拼接方法
|
摘要
近景摄影测量技术在高速铁路轨道几何状态检测中具有较大的应用潜力,而轨道数字影像的准确匹配是图像定向建模的关键环节。本文针对高速铁路轨道近景影像纹理特征差异小且灰度变化不显著的问题,提出采用ORB算法对轨道近景影像进行特征点检测,以最近邻距离与次近邻距离的比值及RANSAC方法完成同名点的匹配,并以匹配的同名点为基础进行相邻影像的拼接。通过在杭甬客运专线上采集的无砟轨道近景影像进行试验,并与常规的SURF算法进行对比分析,结果表明,ORB算法在影像灰度信息高相似性的情况下,能够检测到足够数量且分布均匀的同名点,图像拼接的结果没有缝隙,算法的性能和效率均优于SURF算法,可为近景摄影测量检测高速铁路轨道几何平顺性提供重要的图像技术支撑。
The close-range photogrammetry has high application potential in the detection of the geometric regularity of high-speed railway track.An accurate coregistration of digital track images is the key step of track image oriented and modeling.Considering the minor difference in the texture features and insignificant gray variances of close-range images for high-speed railway track,this paper adopted the ORB algorithm to detect the feature points of the close range track images.Image matching was conducted based on the ratio of the nearest neighbor distance and the second nearest neighbor distance and RANSAC method.Adjacent image mosaic was carried out based on the homologous points after image matching.The proposed method and ORB algorithm were tested using the close-range track images of Hangzhou-Ningbo high-speed railway and were compared with the conventional SURF algorithm.The test results showed that,in the case of high gray similarity of the images,the ORB algorithm can obtain sufficient quality homologous points,as well as seamless mosaic images.The ORB algorithm is better than SURF algorithm in performance and efficiency.The proposed method can provide technological support for the examination of the static geometric regularity of high-speed railway tracks using close-rang photogrammetry.
The close-range photogrammetry has high application potential in the detection of the geometric regularity of high-speed railway track.An accurate coregistration of digital track images is the key step of track image oriented and modeling.Considering the minor difference in the texture features and insignificant gray variances of close-range images for high-speed railway track,this paper adopted the ORB algorithm to detect the feature points of the close range track images.Image matching was conducted based on the ratio of the nearest neighbor distance and the second nearest neighbor distance and RANSAC method.Adjacent image mosaic was carried out based on the homologous points after image matching.The proposed method and ORB algorithm were tested using the close-range track images of Hangzhou-Ningbo high-speed railway and were compared with the conventional SURF algorithm.The test results showed that,in the case of high gray similarity of the images,the ORB algorithm can obtain sufficient quality homologous points,as well as seamless mosaic images.The ORB algorithm is better than SURF algorithm in performance and efficiency.The proposed method can provide technological support for the examination of the static geometric regularity of high-speed railway tracks using close-rang photogrammetry.
引文
[1]熊卫东,周清跃,穆恩生.高速铁路钢轨的平顺性[J].中国铁道科学,2000,21(3):76-83.XIONG Weidong,ZHOU Qingyue,MU Ensheng.The Smoothness of High-speed Railway Track Rails[J].China Railway Science,2000,21(3):76-83.
[2]蔡成标,翟婉明,王其昌.轨道几何平顺性安全限值的研究[J].铁道学报,1995,17(4):82-87.CAI Chengbiao,ZHAI Wanming,WANG Qichang.Study on Allowable Safety Criterion of Track Geometric Irregularities[J].Journal of the China Railway Society,1995,17(4):82-87.
[3]陈强,刘丽瑶,杨莹辉,等.基于双向近景摄影测量检测轨道平顺度的计算模型[J].铁道学报,2012,34(12):83-89.CHEN Qiang,LIU Liyao,YANG Yinghui,et al.Track Regularity Determination Model form Two-way Closerange Photogrammetry[J].Journal of the China Railway Society,2012,34(12):83-89.
[4]陈强,刘丽瑶,杨莹辉,等.高速铁路轨道几何状态的车载摄影快速检测方法与试验[J].铁道学报,2014,36(3):80-86.CHEN Qiang,LIU Liyao,YANG Yinghui,et al.Static Geometry Measurement of High-speed Railway Tracks by Vehicle-borne Photogrammetry[J].Journal of the China Railway Society,2014,36(3):80-86.
[5]张祖勋,张剑清.数字摄影测量学[M].武汉:武汉大学出版社,1997.
[6]RUBLEE E,RABAUD V,KONOLIGE K,et al.ORB:an Efficient Alternative to SIFT or SURF[C]//Proceedings of IEEE International Conference on Computer Vision(ICCV2011).New York:IEEE,2011:2564-2571.
[7]ROSTEN E,DRUMMOND T.Machine Learning for High-speed Corner Detection[C]//Proceedings of the2006European Conference on Computer Vision(ECCV).Berlin Heidelberg:Springer Verlag,2006:430-443.
[8]HARRIS C,STEPHENS M.A Combined Corner and Edge Detector[C]//Proceedings of Alvey Vision Conference.Manchester:AVC,1988:15-50.
[9]ROSIN P L.Measuring Corner Properties[J].Computer Vision and Image Understanding,1999,73(2):291-307.
[10]HU M K.Visual Pattern Recognition by Moment Invariants[J].IRE Transactions on Information Theory,1962,8(2):179-187.
[11]王耀明.图像的矩函数:原理、算法及应用[M].上海:华东理工大学出版社,2002.
[12]CALONDER M,LEPETIT V,STRECHA C,et al.BRIEF:Binary Robust Independent Elementary Features[C]//Proceedings of the 2010European Conference on Computer Vision(ECCV).Berlin Heidelberg:Springer Verlag,2010:778-792.
[13]SZELISKI R.Computer Vision:Algorithms and Applications[M].London:Springer Verlag,2011.
[14]吴福朝.计算机视觉中的数学方法[M].北京:科学出版社,2008.
[15]BAY H,TUYTELAARS T,VAN GOOL L.SURF:Speeded Up Robust Features[C]//Proceedings of the2006European Conference on Computer Vision(ECCV).Berlin Heidelberg:Springer Verlag,2006:404-417.
[2]蔡成标,翟婉明,王其昌.轨道几何平顺性安全限值的研究[J].铁道学报,1995,17(4):82-87.CAI Chengbiao,ZHAI Wanming,WANG Qichang.Study on Allowable Safety Criterion of Track Geometric Irregularities[J].Journal of the China Railway Society,1995,17(4):82-87.
[3]陈强,刘丽瑶,杨莹辉,等.基于双向近景摄影测量检测轨道平顺度的计算模型[J].铁道学报,2012,34(12):83-89.CHEN Qiang,LIU Liyao,YANG Yinghui,et al.Track Regularity Determination Model form Two-way Closerange Photogrammetry[J].Journal of the China Railway Society,2012,34(12):83-89.
[4]陈强,刘丽瑶,杨莹辉,等.高速铁路轨道几何状态的车载摄影快速检测方法与试验[J].铁道学报,2014,36(3):80-86.CHEN Qiang,LIU Liyao,YANG Yinghui,et al.Static Geometry Measurement of High-speed Railway Tracks by Vehicle-borne Photogrammetry[J].Journal of the China Railway Society,2014,36(3):80-86.
[5]张祖勋,张剑清.数字摄影测量学[M].武汉:武汉大学出版社,1997.
[6]RUBLEE E,RABAUD V,KONOLIGE K,et al.ORB:an Efficient Alternative to SIFT or SURF[C]//Proceedings of IEEE International Conference on Computer Vision(ICCV2011).New York:IEEE,2011:2564-2571.
[7]ROSTEN E,DRUMMOND T.Machine Learning for High-speed Corner Detection[C]//Proceedings of the2006European Conference on Computer Vision(ECCV).Berlin Heidelberg:Springer Verlag,2006:430-443.
[8]HARRIS C,STEPHENS M.A Combined Corner and Edge Detector[C]//Proceedings of Alvey Vision Conference.Manchester:AVC,1988:15-50.
[9]ROSIN P L.Measuring Corner Properties[J].Computer Vision and Image Understanding,1999,73(2):291-307.
[10]HU M K.Visual Pattern Recognition by Moment Invariants[J].IRE Transactions on Information Theory,1962,8(2):179-187.
[11]王耀明.图像的矩函数:原理、算法及应用[M].上海:华东理工大学出版社,2002.
[12]CALONDER M,LEPETIT V,STRECHA C,et al.BRIEF:Binary Robust Independent Elementary Features[C]//Proceedings of the 2010European Conference on Computer Vision(ECCV).Berlin Heidelberg:Springer Verlag,2010:778-792.
[13]SZELISKI R.Computer Vision:Algorithms and Applications[M].London:Springer Verlag,2011.
[14]吴福朝.计算机视觉中的数学方法[M].北京:科学出版社,2008.
[15]BAY H,TUYTELAARS T,VAN GOOL L.SURF:Speeded Up Robust Features[C]//Proceedings of the2006European Conference on Computer Vision(ECCV).Berlin Heidelberg:Springer Verlag,2006:404-417.