对偶Voronoi聚类与重网格化  

Dual Voronoi Clustering and Remeshing

作　　者：韩英华[1] 李桂清[1] 彭莉[1] 任灿江[1] 

机构地区：[1]华南理工大学计算机科学与工程学院,广州510640

出　　处：《计算机辅助设计与图形学学报》2009年第11期1535-1544,共10页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家自然科学基金(60673005);广东省自然科学基金(05006540);浙江大学CAD&CG国家重点实验室开放基金

摘　　要：为了以更快的速度得到高质量的多分辨率网格,提出一种基于Voronoi-Delaunay三角化技术的多分辨率表示生成算法.该算法将原三角网格转化为对偶多边形网格再进行Voronoi划分,以自动满足共点聚类块不能超过3个这一约束;根据曲率分布情况来选取基点,以便能更好地捕捉几何特征;最后利用Loop细分规则与局部Laplace平滑指导参数域上的重采样,再映射回模型空间获取最终采样结果,以提高重采样质量.由于Voronoi划分是重网格化算法的瓶颈,采用文中算法能减少划分时条件检测的耗时,从而显著地降低整个重网格化算法的时间复杂度.This paper presents an efficient algorithm for generating multiresolution representations of higher quality by employing Voronoi-Delaunay triangulation. It clusters Voronoi regions on dual polygonal meshes and therefore automatically satisfies the constraint that no more than three Voronoi tiles to share a corner. In addition, it also selects sites under the guidance of curvature distribution in order to capture the geometric features of 3D models. Finally, a resampling strategy combining Loop subdivision and Laplacian smoothing is introduced to enhance the quality of remeshing results. As Voronoi partition is the bottleneck of the algorithm, the adoption of dual polygonal meshes substantially reduces the time for checking the validity of Voronoi partition, hence the algorithm＇s efficiency.

关 键 词：多分辨率 对偶片 重网格化 细分 LAPLACE算子 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]