摘要
提出了一种新的多视点云拼接算法.该方法首先根据点云观测数据,以一个二维平面网格以及网格中各网点所对应的Z坐标来拟合参考点云;然后采用公共点对法将目标点云概略拼合到参考网格点云;最后,计算目标点云中各点至网格点云中相应网眼的距离,并以所有这些距离的平方和最小为准则,采用最小二乘法迭代解算点云坐标转换参数.实例证明,该方法算法可靠,操作简单,计算效率高,具有很强的实用价值.
One novel method for multi-view point cloud registration is presented in this paper. First, an interpolated grid is used to substitute the origin reference point cloud. Then, the object point cloud is aligned to the reference grid point cloud roughly by using common point pairs method. Third, the distances between each point in object point cloud and the related mesh in grid point cloud are calculated. Finally, the coordinate transformation parameters of point cloud are computed with least square method under the guide line that ensures the sum of the distances＇ square is minimum. The method proves reliable, efficient and easy to use.
出处
《同济大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2006年第7期913-918,共6页
Journal of Tongji University:Natural Science
关键词
激光扫描
多视点云
多视拼接
网格点云
laser scanning
multi-view point cloud
multi-view registration
grid point cloud
