Dense flow visualization using wave interference

Abstract

Dense flow visualization and streamlines are among the most popular methods in numerous applications of scientific visualization. Because very few works synthesize the benefits of these approaches into one framework, we propose a hybrid technique. In this work, we present a novel method of dense flow visualization that produces high contrast images (typical of streamline approaches), with manageable spatial frequency.We begin by introducing a notion of a streamline thickness function and thickness wave sources, and then compute the resulting image as a coherent wave interference pattern. We show that this problem is equivalent to finding the dominant eigenvector of a sparse matrix and employ the power iterations numerical scheme to demonstrate an efficient parallel implementation. We conclude by discussing possible quality improvement strategies and extensions of our method to other domains, including 3D flows.

A Discrete Probabilistic Approach to Dense Flow Visualization

“…The technical side of our approach is inspired by the wave interference method in DFV [39], relying on the similar mathematical apparatus of sparse matrix computations. The eigenvector computation, resulting from the analysis of our probabilistic model, has a direct correspondence to the spectral embeddings technique widely used in the image processing domain.…”