Topology based Tensor Field Segmentation
The goal of this effort is to develop a method for computing a tensor field segmentation based on the topological structure of the eigenvector-fields and the corresponding eigenvalue-fields. For the extraction of the topological skeleton we use eigenvector based interpolation schema making it very simple. The eigenvalue field is used to guide a subdivision and simplification of the topological skeleton into regions of similar behavior. This segmentation allows one to comprehend a tensor field's structure on a qualitative level, and it also makes it possible to determine regions of similar (or dissimilar) behavior. The resulting higher-level characterizations of a tensor field provide valuable information to analyze field behavior on a more abstract level. Our research is motivated by applications encountered in areas like civil engineering and geophysics, where computer simulations often produce tensor field data for which appropriate tools for interpretation are still in their infancy. We have applied our method to a data set representing the simulation of a simple two-point load on a solid block to evaluate its correctness, efficiency, and potential for topology-based analysis of complex tensor fields. Further information is available in the detailed project description.