文章摘要
引用本文:张伯雄,谢伙生.基于特征分布的三维流线相似性研究[J].福州大学学报(自然科学版),2016,44(5):633~638
基于特征分布的三维流线相似性研究
3D stremlines similarity analysis based on distribution of measurements
  
DOI:10.7631/issn.1000-2243.2016.05.0633
中文关键词: 流线相似性  特征分布  二维直方图  流线聚类
英文关键词: streamline similarity  feature distribution  2D histogram  streamline clustering
基金项目:
作者单位
张伯雄 福州大学数学与计算机科学学院福建 福州 350116 
谢伙生 福州大学数学与计算机科学学院福建 福州 350116 
摘要点击次数: 233
全文下载次数: 205
中文摘要:
      针对利用流线的形状特征对流线进行分类和选取,可以方便用户洞察三维流场的特征,提出一种高效的基于特征分布的三维流线相似性比较方法. 该方法在Lu Kewei等流线相似性比较方法的基础上,引入曲折度和速度方向熵两个全局属性. 首先将流线均匀分段,然后计算每段的曲率直方图、扭率直方图、曲折度直方图、速度方向熵直方图,构建相应的二维直方图,最后利用堆土机距离(EMD)及k-means聚类方法进行流线相似度计算和分类. 实验结果表明,该方法在引入全局几何属性后能够产生更鲁棒性的查询和分类结果.
英文摘要:
      Considering the feature of shape,classification and selection of streamlines advance the understanding of the features in 3D flow field.In this paper,we propose an efficient 3D streamlines similarity comparison method based on distribution of features. This method is based on the streamline similarity comparison method proposed by Lu Kewei et al. and introduces two global geometric properties:tortuosity and velocity direction entropy. At first,we divide each streamline into segments evenly,then we construct curvature histogram、torsion histogram、tortuosity histogram and velocity direction entropy histogram for each segment,and form corresponding 2D histograms.In the end,we use the Earth Mover’s Distance(EMD) and k-means clustering method to measure the similarity between streamlines and classify these streamlines. The final experiment showed that after combining the two global features,this method can produce more robust query and clustering results.
查看全文   查看/发表评论  下载PDF阅读器
关闭