面向大规模体数据集的复杂几何曲面抽取方法  

A Complex Geometric Surface Extracting Method for Large-Scale Volumetric Data

作　　者：吴付坤 肖丽[2] 王华维[2] Wu Fukun;Xiao Li;Wang Huawei(Software Center for High Performance Numerical Simulation,China Academy of Engineering Physics,Beijing 100088;Institute of Applied Physics and Computational Mathematics,Beijing 100088)

机构地区：[1]中物院高性能数值模拟软件中心,北京100088 [2]北京应用物理与计算数学研究所,北京100088

出　　处：《计算机辅助设计与图形学学报》2019年第11期1908-1916,共9页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家重点研发计划(2017YFB0202203);中国工程物理研究院创新发展基金(PY2019126)

摘　　要：针对在大规模数据场中进行复杂几何曲面抽取的性能和精度问题,提出一种基于八叉树加速结构和多边形裁剪的几何曲面抽取方法.首先应用平面-体求交方程将体单元抽取问题转换到二维空间中,其中引入基于空间包围盒的八叉树结构以减少无效体单元的计算;然后提出面向凹凸多边形的裁剪方法和若干优化策略以有效地进行多边形裁剪计算;再引入数据集预筛选功能减少无效网格片读入以提升曲面抽取性能;最后将该方法集成到通用可视分析引擎中.实验结果表明,该方法能够高精度、高性能地从数值模拟程序生成的数据场中抽取复杂几何曲面的表面场值.To address the performance and precision problem of extracting volumetric data using complex geometric surface, we present and implement an efficient surface extraction method based on octree acceleration structure and polygon clipping for visualizing significant field data on arbitrary object surface in 3 D space. This method first applies plane-volume intersection equation to transform the extraction problem of volume cells to 2 D space where an octree acceleration structure based on spatial bounding box is novelly introduced. Then, a clipping algorithm for concave and convex polygons and several optimizations are specifically designed to efficiently calculate intersection polygon. Pre-selection strategy is introduced to dramatically reduce the number of invalid mesh patches in order to improve performance. Finally, this method is integrated into common visual analysis engine. Our experiments show that the proposed method can achieve the surface extraction operation of complex geometry from volumetric dataset generated by numerical simulation with high precision and performance.

关 键 词：几何曲面抽取 平面-体求交方程 八叉树加速结构 多边形裁剪 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]