Superresolution and Denoising of 3D Fluid Flow Estimates

  • Authors:
  • Andrey Vlasenko;Christoph Schnörr

  • Affiliations:
  • Image and Pattern Analysis Group, Heidelberg Collaboratory for Image Proc.(HCI), University of Heidelberg, Germany;Image and Pattern Analysis Group, Heidelberg Collaboratory for Image Proc.(HCI), University of Heidelberg, Germany

  • Venue:
  • Proceedings of the 31st DAGM Symposium on Pattern Recognition
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

Three-dimensional high-speed image measurements of fluid flows become state-of-the-art velocimetry techniques in experimental fluid mechanics and related areas of industry. In this paper, we consider data produced by an established technique, sparse Particle Tracking Velocimetry (PTV), in terms of small set of velocity estimates irregularly distributed over a 3D volume, and study a variational approach to super-resolution in a physically consistent way. The output consists of a high-resolution vector field on the voxel grid where additionally the typical quantization noise of the input data can be removed. Numerical experiments validate our approach with 3D data from turbulent flows.