中文摘要：

为了有效地解决三维流场可视化中由于播种流线所产生的众所周知的遮挡和杂乱问题，呈现出清晰的流场模式与流场的重要特征，提出一种基于迭代最邻近点（ICP）与K均值聚类的流线提取算法．首先利用ICP算法实现流线间轮廓特征上的配准，并根据几何相似性进行排序；然后利用K均值聚类算法对流线分组；最后根据用户指定密度约简多余相似性流线，并以此结果重构矢量场来评价文中算法的精确度．将文中算法应用到多个数据集进行实验并与已有的流线分布的最新算法进行比较，结果表明，该算法能更有效地反映流场的关键特性，大大提高了三维流场数据集的可读性．

英文摘要：

This paper addresses the well known occlusion and cluttering issue effectively in the visualization of 3D flow fields using streamlines. A streamline selection algorithm based on iterative closest point （ICP） and K-means clustering is presented to show a clear flow field pattern and the important flow features. The ICP method registers shape and ranks streamlines, and then the ranking values are used to cluster the initial streamlines. Finally, the similar streamlines in the same cluster are pruned according to the density defined arbitrarily by the user. An effectiveness estimation function is also introduced to evaluate the accuracy of the presented algorithm with the reconstruction of the vector field. Compared with previous work in uniform streamline placement, comprehensive experiments conducted on multiple datasets show that the proposed algorithm can better reflect the underlying properties of the flow field and greatly improve the readability of 3D flow field datasets.