中文摘要：

基于流线的矢量场可视化是科学可视化的重要分支,在许多领域都有重要应用。但是现有的可视化方法很难有效评估流线对矢量场信息的反映程度。为了解决这一问题,提出了一种基于信息熵的矢量场种子点评估和布局算法。首先,基于信息熵进行初始布点捕捉矢量场中的主要特征,并利用模板将这些特征加强显示。然后,基于生成的流线恢复中间矢量场,通过恢复的矢量场和输入矢量场的条件熵值量化评估现有流线对矢量场的信息反映程度。根据条件熵和重要性添加新种子点。重新评估流线质量,反复迭代直到条件熵收敛。实验表明,该方法可以用较少的流线反映出矢量场中较多的信息。

英文摘要：

Flow visualization based on streamlines, which is applied in many fields, is a very important branch of visualization. However it’s difficult to effectively measure how much information of the origin flow field have been displayed by streamlines generated with existing methods. A new method which can evaluate streamlines and place seeds based-on information entropy is presented in this paper. Firstly, the seeds are placed based on Shannon entropy which can seize the important features in the field, and then the features are strengthened by templates. The conditional entropy between intermediate flow field and origin vector field are used to quantify the uncertainty remained in original field after streamlines are shown. An importance-based seed sampling method adds new seeds iteratively until the conditional entropy between intermediate flow field and origin vector field converges. Experiments show that this algorithm can show enough information to recover the origin field with limited streamlines.