Multitype Multifield Visualization

多场数据的可视化分析是科学可视化领域的重大研究挑战之一。近年来，已经提出了许多方法，这些方法要么对字段进行并行可视化，要么采用半自动方法计算和可视化字段之间的关系。然而，大多数现有技术都关注于同一类型的多字段，例如多个标量字段的集合。最近的多场研究趋向于由不同类型的场组成，即在同一域中获取标量场、矢量场和张量场。本项目提出了不同类型的多场可视化的方法。特别地，我们提出了多类型字段之间的相似度量，定义了描述多类型字段之间相关性的特征，并将信息可视化中的离散方法应用于多类型字段。这些方法在来自不同应用领域的大量测试数据集上进行了测试和评估。

The visual analysis of multifield data is one of the big research challenges in the field of Scientific Visualization. In recent years, many approaches for this have been proposed which either do a side-by-side visualization if the fields or apply semi-automatic methods to compute and visualize the relations between the fields. However, most existing techniques focus on multifields of the same type, for instance a collection of multiple scalar fields. Recent multifields tend to consist of fields of different types, i.e., scalar, vector andtensor fields are acquired over the same domain. This project proposes approaches for the visualization of multifields of different types. In particular, we propose similarity measures between multitype fields, we define features describing the correlations between multitype fields, and we adapt discrete methods from Information Visualization for multitype fields. The approaches are tested and evaluated on a number of test data sets from different areas of application.