期刊文献+

一种新的针对中国古建筑室内三维激光扫描数据的配准方法 被引量:4

A new registration method aiming at 3D laser scanning data of inner old China architecture
下载PDF
导出
摘要 针对中国古建筑结构复杂、室内古建筑激光扫描数据量大、视点多、不易采用现有的配准方法等问题,提出基于古建筑按中轴线对称等几何结构特征的古建筑室内激光扫描数据配准方法,同时首次提出广义同名点的概念。建立测站坐标系、建筑坐标系和测站辅助坐标系,并详细介绍从测站直角坐标系转换到测站辅助坐标系的方法,同时给出具体的坐标转换关系式。该配准方法分两步进行:第一步,将各视点扫描数据从测站直角坐标系转换到各自的测站辅助坐标系;第二步,选定一测站辅助坐标系为参考坐标系,之后利用广义同名点求得辅助坐标系之间的三维平移向量,最终实现配准。该方法不需初始位姿,计算量小,不易受噪声影响,实验结果验证了这种配准方法的快速、高效性。 This paper presents a new registration method aiming at 3D laser scanning data of inner old China architecture based on the characteristics of geometry structure of old architceture such as symmetry by middle axes, in order to solve the question that common registration methods are not well applied because of large volume data, multi - viewpoints and complicated structure of old arehitecturc. The concept of generalized matching point is put forward. After building viewpoint coordinate system(VCS), architecture coordinate system (ACS) and viewpoint assistant coordinate system(VACS), the paper intreduces how to transform from VCS to VACS, and gives the rigorous coordinate transformation relatianship equations. During the process of registration, every viewpoint scanning data are first transformed from VCS to VACS, and then a VCS selected is taken as reference coordinate system, and then generalized matching points are utilized to compute the 3D transformation vector. Thus registration is finished. This methed with fast computing speeding does not need initial pose estimation, and is less affected by noise. Experiment results show its reliability and robustness.
出处 《激光杂志》 CAS CSCD 北大核心 2006年第6期63-65,共3页 Laser Journal
基金 武汉大学测绘遥感信息工程国家重点实验室开放基金资助项目(编号:WKL(04)0102)
关键词 配准 古建筑 三维激光扫描数据 坐标转换 广义同名点 registration old architecture 3D Laser scanning data coordinate transformatiopn generalized matching point
  • 相关文献

参考文献8

  • 1P.J.Besl,N.D.McKay.A method for registration of 3-D shapes[J].IEEE Trans.Pattern Anal.Mach.Intell.,1992,14 (2):239-256. 被引量:1
  • 2Z.Zhang.Iterative point matching for registration of free-form curves and surfaces[J].Int.J.Comput.Vision,1994,13 (2):119-152. 被引量:1
  • 3M.Greenspan,M.Yurick.Approximate K-D tree search for efficient ICP[C].in:Proc.4th Internat.Conf.on 3D Digital Imaging and Modeling,2003,442-448. 被引量:1
  • 4T.Jost,H.Hugli.A multi-resolution ICP with heuristic closest point search for fast and robust 3D registration of range images[C].in:Proc.4th Internat.Conf.on 3D Digital Imaging and Modeling,2003,427-433. 被引量:1
  • 5C.Langis,M.Greenspan,G.Godin.The parallel iterative closest point algorithm[C].in:Proc.3rd Internat.Conf.on 3D Digital Imaging and Modeling,2001,195-202. 被引量:1
  • 6J.P.Thirion.Extremal points:definition and application for 3D image registration[C].in:Proc.IEEE Internat.Conf.on Computer Vision and Pattern Recognition,1994,587-592 被引量:1
  • 7P.Krsek,T.Pajdla,V.Hlavac.Differential invariants as the base of triangulated surface registration[J].Comput.Vision Image Und.,2002,87:27-38. 被引量:1
  • 8Huber,D.,Carmichael,O.,and Hebert,M.3D map reconstruction from range data[C].in:Proc.of IEEE ICRA,2000,891-897. 被引量:1

同被引文献34

引证文献4

二级引证文献29

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部