A viewer-dependent tensor field visualization using particle tracing

Abstract

Tensor field visualization is a hard task due to the multivariate data contained in each local tensor. In this paper, we propose a particle-tracing strategy to let the observer understand the field singularities. Our method is a viewer-dependent approach that induces the human perceptual system to notice underlying structures of the tensor field. Particles move throughout the field in function of anisotropic features of local tensors. We propose a easy to compute, viewer-dependent, priority list representing the best locations in tensor field for creating new particles. Our results show that our method is suitable for positive semi-definite tensor fields representing distinct objects. © 2011 Springer-Verlag.