利用球目标的多测站点云自动配准
|
摘要
提出了一种基于球心点互斥的球目标识别方法,并将其应用于多测站点云的自动配准。首先,针对每一个测站点云,根据法向和曲率推算所有可能出现球目标的球心点位置;然后,设计了一种仅利用球心点之间距离和局部密度的球心点互斥算法,用来探测和定位杂乱场景中的球目标,并利用球面覆盖率指标评价球目标的可靠性;最后,根据成功匹配的可靠球目标整体配准各个测站。针对由多种几何体搭建的实验场景从4个不同测站采集三维激光扫描点云,单点定位误差大于3mm。实验结果表明,在未知球目标数量、半径且存在遮挡的情况下,能够准确识别出可靠的球目标,并利用同名目标完成不同测站间的精确配准,配准后球心残差小于1mm。
A method of Spherical Targets recongnition is designed based on sphere center mutual exclusion and is used to registrate automaticly multi-station point clouds.First,for each station point clound,we estimate all possible spherical centers according to the normal and curvature,then design an algorithm of sphere center mutual exclusion which exploits the distance between spherical center points and their local densities to probe and locate spherical targets in cluttered scenes.And then,the coverage ratio of spherical points is used to evaluate the reliability of these possible spheres,which are subsequently used to register multiple stations.Four stations of point clouds acquired from a cluttered scene composed of many types of geometric models are used to verify the effectiveness of the supported method.Experimental process demonstrates the robustness of spheres recognition and the accuracy of center positioning.The final registration results show that every sphere residual error is less than 1 mm.
A method of Spherical Targets recongnition is designed based on sphere center mutual exclusion and is used to registrate automaticly multi-station point clouds.First,for each station point clound,we estimate all possible spherical centers according to the normal and curvature,then design an algorithm of sphere center mutual exclusion which exploits the distance between spherical center points and their local densities to probe and locate spherical targets in cluttered scenes.And then,the coverage ratio of spherical points is used to evaluate the reliability of these possible spheres,which are subsequently used to register multiple stations.Four stations of point clouds acquired from a cluttered scene composed of many types of geometric models are used to verify the effectiveness of the supported method.Experimental process demonstrates the robustness of spheres recognition and the accuracy of center positioning.The final registration results show that every sphere residual error is less than 1 mm.
引文
[1]张毅,闫利,杨红,等.地面激光扫描球形标靶的球心误差研究[J].武汉大学学报·信息科学版,2012,37(5):598-601
[2]郭敬平.基于探测球的固定式扫描海量点云自动定向方法[J].测绘工程,2015,24(10):11-14
[3]张学昌,唐艳梅,梁涛.基于球面特征的点云配准方法研究[J].机械科学与技术,2015,34(12):1 851-1 856
[4] Franaszek M,Cheok G S,Witzgal C.Fast Automatic Registration of Range Images from 3DImaging Systems Using Sphere Targets[J].Automation in Construction,2009,18(3):265-274
[5] Wang Y M,Shi H B,Zhang Y Y,et al.Automatic Registration of Laser Point Cloud Using Precisely Located Sphere Targets[J].Journal of Applied Remote Sensing,2014,8(1):5 230-5 237
[6] Cazals F,Pouget M.Estimating Differential Quantities Using Polynomial Fitting of Osculating Jets[J].Computer Aided Geometric Design,2005,22(2):121-146
[7]李宏博,史先琦,万奇灵,等.GPS高程拟合在丘陵地区地形测量中的应用研究[J].测绘地理信息,2015,40(4):44-47
[8] Carmo M P D.Differential Geometry of Curves and Surfaces[M].Englewood Cliffs:Prentice Hall,1976
[9]林良彬,杨必胜,代文霞,等.多视角林地地面激光点云自动化配准[J].测绘地理信息,2015,40(6):86-89
[10]潘国荣,范伟.船舶测量自由移站法及多站转换参数的整体平差[J].山东科技大学学报(自然科学版),2013,32(3):64-70
[11]郑德华.ICP算法及其在建筑物扫描点云数据配准中的应用[J].测绘科学,2007,32(2):31-32
[2]郭敬平.基于探测球的固定式扫描海量点云自动定向方法[J].测绘工程,2015,24(10):11-14
[3]张学昌,唐艳梅,梁涛.基于球面特征的点云配准方法研究[J].机械科学与技术,2015,34(12):1 851-1 856
[4] Franaszek M,Cheok G S,Witzgal C.Fast Automatic Registration of Range Images from 3DImaging Systems Using Sphere Targets[J].Automation in Construction,2009,18(3):265-274
[5] Wang Y M,Shi H B,Zhang Y Y,et al.Automatic Registration of Laser Point Cloud Using Precisely Located Sphere Targets[J].Journal of Applied Remote Sensing,2014,8(1):5 230-5 237
[6] Cazals F,Pouget M.Estimating Differential Quantities Using Polynomial Fitting of Osculating Jets[J].Computer Aided Geometric Design,2005,22(2):121-146
[7]李宏博,史先琦,万奇灵,等.GPS高程拟合在丘陵地区地形测量中的应用研究[J].测绘地理信息,2015,40(4):44-47
[8] Carmo M P D.Differential Geometry of Curves and Surfaces[M].Englewood Cliffs:Prentice Hall,1976
[9]林良彬,杨必胜,代文霞,等.多视角林地地面激光点云自动化配准[J].测绘地理信息,2015,40(6):86-89
[10]潘国荣,范伟.船舶测量自由移站法及多站转换参数的整体平差[J].山东科技大学学报(自然科学版),2013,32(3):64-70
[11]郑德华.ICP算法及其在建筑物扫描点云数据配准中的应用[J].测绘科学,2007,32(2):31-32