改进的泊松曲面重建算法  

Improved reconstruction algorithm of Poisson surface

作　　者：曹诗卉[1] 亓迎川[1] 时满宏[1] CAO Shihui;QI Yingchuan,;SHI Manhong(Air Force EarlyWarning Academy,Wuhan 430019, China)

机构地区：[1]空军预警学院,武汉430019

出　　处：《空军预警学院学报》2016年第4期289-291,302,共4页Journal of Air Force Early Warning Academy

摘　　要：针对非均匀采样或有噪声的有向点云,提出了一种改进的泊松曲面重建算法.首先选择一个隐式化参数,利用凸优化取代等几何有限元方法和原始对偶算法来有效地解决算法最小化问题;然后利用凸最小化对未知拓扑结构的点云数据的内曲面的指示函数进行重构;最后对改进算法进行仿真分析.仿真结果表明,改进算法对噪声和异常点具有鲁棒性,所得到的重构曲面光顺性好.This paper proposes an improved Poisson surface reconstruction algorithm for the oriented pointclouds with non-uniform sampling and noise. Firstly, We choose an implicitization parameter, use the convexoptimization to replace the geometrical finite-element method, and employ the primal-dual algorithm to solveeffectively the algorithm minimization; and then, we reconstruct the internal surface’s indicator function of thepoint cloud data with unknown topological structure using the convex minimization, and finally perform asimulated analysis of the improved algorithm. Simulation results show that this improved algorithm is ofrobustness for noise and outliers, and the obtained reconstruction surface is also of better smoothness.

关 键 词：凸最小化 指示函数 鲁棒性 光顺性 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]