This project is concerned with the visualization and analysis of three-dimensional tensor fields. Due to the complexity of tensor data, visualization methods have to be chosen carefully. This is especially true for three-dimensional fields. In a pre-processing step filter methods have to be applied to reduce the field to relevant features. A challenging question in this context is the question: "What are relevant features?".

Tensors provide a powerful mathematical language to describe physical phenomena.
Consequently, they have a long tradition in physics and appear in various application
areas, either as intermediate product or as output of simulations or measurements. Examples are medicine, astrophysics, continuum mechanics, image processing, and many more. The potential of tensors to describe complex anisotropic behavior, however, concurrently complicates their interpretation. Scientific visualization can provide insight into the processes that are described by tensors. The goal of this project is the development of methods that effectively communicate the complex information contained in three-dimensional tensor fields.

The basic idea of our approach is to provide various perspectives onto the data via linking of well-known and novel visualization methods. Therefore, we combine attribute-space views and three-dimensional spatial visualizations (Figure 2). A particular challenge in this context arises if we have no a-priori knowledge about the data. In that case, automatic segmentation algorithms to reduce the data to relevant features are not applicable. In this project, we, therefore, focus on explorative visualization.