摘要
三维激光扫描是近年来兴起的一种全新的测量技术,能快速获取研究物体表面海量的3D坐标信息,即点云。由于三维激光扫描具有数据获取速度快、精度高、不接触物体等优点,因此利用点云进行场景的三维重建成为人们研究的热点,其中点云配准是其中的关键技术之一。本文简述了点云配准的概念,对各种点云配准方法进行了阐述及分析,总结了当前点云配准中存在的问题及进一步的研究方向。
Terrestrial laser scanner is a new and rising surveying technology, with this technology, large number of 3D coordinates of observed object, known as point clouds, can be obtained quickly and accurately. Because of having such advantages, it becomes hot topic for many researchers of 3D modeling to pay attention to. Due to the reasons such as field of view and occlusion, multiple point clouds from different viewpoint should be acquired. But the point coordinates obtained in different viewpoint is relative to its own viewpoint, in order to construct the complete 3D model of investigated object, point clouds data should be transformed into a common reference frame. This process, called registration, is a crucial step in 3D reconstruction. In this paper, the definition of registration is narrated first, followed by reviewing and analyzing of all kinds of registration methods of point clouds. The problems and further study of registration for point clouds is pointed out.
出处
《工程勘察》
CSCD
北大核心
2008年第1期53-57,共5页
Geotechnical Investigation & Surveying
基金
国家自然科学基金资助项目(40574008)
江西省教育厅科技项目(赣教技字[2007]237号)
江西省科技攻关项目.
关键词
三维激光扫描
点云
点云配准
最邻近点迭代法(ICP)
3D laser scanning
point clouds
point clouds registration
iterative closest point algorithm
