Visual Analysis of Spatia-temporal Relations of Pairwise Attributes in Unsteady Flow
Visual Analysis of Spatia-temporal Relations of Pairwise Attributes in Unsteady Flow
复制标题
DOI：
10.1109/tvcg.2018.2864817
复制
发表时间：
2019-01
影响因子：
5.2
通讯作者：
Marzieh Berenjkoub;R. O. Monico;R. Laramee;Guoning Chen
中科院分区：
文献类型：
--
作者：
Marzieh Berenjkoub;R. O. Monico;R. Laramee;Guoning Chen
尽管在非定常流动的分析和可视化方面取得了重大进展，但对其行为的解释仍然是一个挑战。在这项工作中，我们重点研究了非定常流动不同物理属性之间的线性相关性和非线性相关性，以期从一个新的角度对其进行研究。具体地说，我们将现有的空间相关性量化，即局部相关系数(LCC)扩展到时空域，从欧拉和拉格朗日两个角度研究属性对的相关性。为了研究属性之间的依赖关系，不一定是线性的，我们扩展并计算了属性之间的互信息。为了帮助可视化和解释与粒子相关联的属性之间派生的相关性和依赖性，我们对单个路径上的相关性和依赖性值进行编码。最后，为了利用相关性和MI计算结果来识别具有感兴趣流量行为的区域，我们提出了一种基于属性关系强度排序的流量域分割策略。我们已经将我们的相关性和相依性度量应用于具有不同时空核大小的2D和3D非定常流动，以演示和评估它们的有效性。
Despite significant advances in the analysis and visualization of unsteady flow, the interpretation of it's behavior still remains a challenge. In this work, we focus on the linear correlation and non-linear dependency of different physical attributes of unsteady flows to aid their study from a new perspective. Specifically, we extend the existing spatial correlation quantification, i.e. the Local Correlation Coefficient (LCC), to the spatio-temporal domain to study the correlation of attribute-pairs from both the Eulerian and Lagrangian views. To study the dependency among attributes, which need not be linear, we extend and compute the mutual information (MI) among attributes over time. To help visualize and interpret the derived correlation and dependency among attributes associated with a particle, we encode the correlation and dependency values on individual pathlines. Finally, to utilize the correlation and MI computation results to identify regions with interesting flow behavior, we propose a segmentation strategy of the flow domain based on the ranking of the strength of the attributes relations. We have applied our correlation and dependency metrics to a number of 2D and 3D unsteady flows with varying spatio-temporal kernel sizes to demonstrate and assess their effectiveness.