高阶平滑表面提取算法的CUDA并行实现

计算机科学 ›› 2013, Vol. 40 ›› Issue (6): 29-31.

高阶平滑表面提取算法的CUDA并行实现

袁红星,吴少群,郭立,朱仁祥

宁波工程学院电子与信息工程学院宁波315016;宁波工程学院电子与信息工程学院宁波315016;中国科学技术大学电子科学与技术系合肥230026;宁波工程学院电子与信息工程学院宁波315016

出版日期:2018-11-16 发布日期:2018-11-16
基金资助:
本文受宁波市自然科学基金(2012A610043),浙江省自然科学基金(LY12F01001),国家自然科学基金(61071173)资助

Higher-order Smooth Surface Extraction CUDA Parallel Implementation

YUAN Hong-xing,WU Shao-qun,GUO Li and ZHU Ren-xiang

Online:2018-11-16 Published:2018-11-16

摘要/Abstract

摘要： 高阶平滑表面提取算法可有效抑制传统步进立方体算法存在的鳞状失真现象,但引入了较复杂的最优化运算,降低了表面提取的效率。针对该问题,提出基于图形处理器的并行加速优化方法。首先将算法分解成分界区域、窄带区域、嵌入函数边界值、嵌入函数值最优化和三角面网格提取5个计算步骤,其次根据每个处理步骤的运算特点进行任务分解。为便于图形处理器并行优化,将其中最耗时的嵌入函数值最优化计算,表示成矩阵运算形式,通过投影雅可比迭代估计最优解。实验结果表明,在GeForce GT 240M显卡上并行优化后平均加速比可达到9以上。

关键词: 步进立方体,表面提取,图形处理器,高阶平滑

Abstract: Higher-order smooth surface extraction method can overcome aliasing artifacts of marching cubes algorithm．However,it introduces extra computation burden especially with optimal embedding function calculation．To resolve the problem,a parallel implementation based on graphics processing unit was presented．The original higher-order smooth surface extraction algorithm was divided into five parts including margin region,narrow band region,embedding function margin values,optimal embedding function and triangular mesh extraction．Then they were paralleled with task assignments method．The most complexity function embedding calculation is approximated by projected Jacobin method．The experimental results show that the speedup achieves more than 9after parallelization on GeForce GT 240M GPU.

Key words: Marching cubes,Surface extraction,Graphics processing unit,Higher-order smooth

袁红星,吴少群,郭立,朱仁祥. 高阶平滑表面提取算法的CUDA并行实现[J]. 计算机科学, 2013, 40(6): 29-31. https://doi.org/

YUAN Hong-xing,WU Shao-qun,GUO Li and ZHU Ren-xiang. Higher-order Smooth Surface Extraction CUDA Parallel Implementation[J]. Computer Science, 2013, 40(6): 29-31. https://doi.org/

参考文献

[1] 僧德文,李仲学,李翠平,等．Marching Cubes算法改进研究及应用[J]．计算机应用研究,2006,7:50-51
[2] 梁秀霞,张彩明．拓扑结构正确的三线性插值曲面的三角片逼近[J]．计算机研究与发展,2006,43(3):528-535
[3] Lopes A,Brodlie K．Improving the robustness and accuracy of the marching cubes algorithm for isosurfacing[J]．IEEE Tran-sactions on Visualization and Computer Graphics,2003,9(1):16-29
[4] 薛强,蔡文立,石教英．Marching Boxes:一个多精度等值面抽取算法[J].计算机辅助设计与图形学学报,1998,10(1):7-14
[5] 朱经纬,蒙培生,王乘．一种改进的MC算法[J]．中国图象图形学报,2008,13(7):1359-1366
[6] 吕理伟,顾耀林．移动立法体算法的三重线性插值研究[J]．计算机工程与应用,2005,32:41-44
[7] Lempitsky V,Boykov Y．Global optimization for shape fitting[C]∥Proceedings of Computer Vision and Pattern Recognition,2007．Minneapolis,USA:IEEE Press,2007:1-8
[8] Li Wen,Guo Li,Yuan Hong-xing,et al．Parallel implementation and optimization of the Sebvhos algorithm[J]．Journal of Electronics,2011,28(3):277-283
[9] 李文,郭立,袁红星,等．一种高阶平滑表面并行提取方法[J]．中国科学院研究生院学报,2012,29(2):251-256
[10] NVIDIA Corporation．CUDA programming guide 4.0 [EB/OL]．http://www.developer.nvidia.com/,2011-06-27
[11] Cremeters D,Kolev K．Multiview stereo and silhouette consistency via convex functionals over convex domains [J]．IEEE Transactions on pattern analysis and machine intelligence,2011,33(6):1161-1174
[12] NVIDIA Corporation.CUDA SDK C Samples[EB/OL]． http://developer.nvidia.com/cuda-cc-sdk-code-samples,2012-06-27
[13] Culess B,Levoy M．A volumetric method for building complexmodels from range images[C]∥Proceedings of ACM SIGGRAPH,1996．New Orleans,LA,USA:ACM Press,1996:303-312
[14] Yasutaka F,Jean P．3D Photography Dataset [EB/OL]． http://www.cs.washington.edu/homes/furukawa/research/mview/index.html,2011-06-27
[15] Steve S,Brian C,James D,et al．The multi-view stereo evaluation[EB/OL]．2011-06-27

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed