摘要
针对噪声点云曲面重建,提出了一种基于Delaunay细分的曲面重建算法。首先以点云法向为约束,采用抗差估计的方法拟合球面近似局部曲面;然后利用沿坐标轴的包围盒树结构(axis aligned bounding boxes tree,AABB-tree)快速搜索与线段相交的曲面包围球,以各包围球球心为初值、半径为可信区间,并行化迭代计算出线段与球面的首个交点,该交点可近似为线段与曲面交点;最后不断地插入交点进行Delaunay细分,从而网格化曲面。实验结果表明,当点云噪声较大时,该方法可以快速、稳健地重建出高质量曲面,且曲面重建精度较高。
To reconstruct surfaces from noisy point clouds,a surface reconstruction algorithm based on Delaunay refinementis proposed.First,a local surface is approximated by algebraic sphere,fitted through neighboring point coordinates and normals by the robust least square algorithm.Compared to traditional sphere fitting methods,the new method is more robust to noise and outliers.Secondly,to find any segment intersecting a surface in the Delaunay refinement procedure,surface bounding spheres intersecting segments were efficiently located using the AABB-tree algorithm.Then,initialized by the sphere center,initial segment-surface intersections approximated within bounding spheres were parallel-computed by iterative segment-sphere intersection.Finally,the surface was meshed by Delaunay refinement,unambiguously.This approach can reconstruct surfaces with a good aspect ratio in comparison with the matching-cube algorithm.Experiments show that the new algorithm can efficiently,robustly,and accurately reconstruct surfaces from point clouds with high noise,but the time cost and memory consumption rapidly increases when executing precise models.
作者
李国俊
李宗春
孙元超
李伟
黄志勇
LI Guojun LI Zongchun SUN Yuanchao LI Wei HUANG Zhiyong(School of Mapping and Surveying, Information Engineering University, Zhengzhou 450052, China Beijing Satellite Navigation Center, Beijing 100094, China China Aerospace Surveying ＆ Mapping Center, Beijing 102102, China)
出处
《武汉大学学报（信息科学版）》
EI
CSCD
北大核心
2017年第1期123-129,共7页
Geomatics and Information Science of Wuhan University
