Topology Aware Vector Field Denoising

Recent developments in data acquisition technology enable to directly capture real vector fields, helping for a better understanding of physical phenomena. However measured data is corrupted by noise, puzzling the understanding of the phenomena. This turns the task of removing noise, i.e. denoising, an essential preprocessing step for a better analysis of the data. Nonetheless a careful use of denoising is required since usual algorithms not only remove the noise but can also eliminate information, in particular the vector field singularities, which are fundamental features in the analysis. This paper proposes a semi-automatic vector field denoising methodology, where the user visually controls the topological changes caused by classical vector field filtering in scale-spaces.