摘要
提出一种改进的模型表面点云的近似最小包围盒求解方法,该方法首先构建模型表面采样数据的动态空间索引结构,依据点云特征型面的曲率在保持型面特征的基础上对原始数据进行精简,采用均值漂移算法得到新的模式点集进行二次精简,计算精简后数据的凸包,利用O′Rourke提出的凸多面体的最小包围盒求解算法获得凸包的最小包围盒并确定局部坐标系,利用局部坐标系求解原始点云数据的近似最小包围盒,可在满足最小包围盒体积精度的同时提高算法的运行效率,能有效处理各种复杂型面的点云数据的最小包围盒快速求解问题。
An accelerating algorithm of approximate minimum bounding box for sampled data of surface is proposed. Firstly, a dynamic spatial index structure of the scattered points was established with R S-tree. Then the scattered points were streamlined based on curvature of the local surface and the Mean-shift algorithm. The minimum bounding box was solved using the O＇Rourke＇s algorithm with the convex hull of points and the local coordinate system was determined at the same time. The box of the original points was solved in the local coordinate system. It is proved that this method can improve the minimum bounding box efficiently and ensure the accuracy of the scattered points with complex surface.
出处
《农业装备与车辆工程》
2012年第4期1-5,共5页
Agricultural Equipment & Vehicle Engineering
基金
国家自然科学基金资助项目(51075247)
山东省自然科学基金资助项目(ZR2010EM008)
关键词
海量散乱点云
数据精简
曲率
均值漂移
最小包围盒
dense scattered points
data reduction
curvature
mean shift
minimum bounding box
