Comparison among feature tracking and more consolidated velocimetry image analysis techniques in a fully developed turbulent channel flow

Abstract

The presence of a large number of software codes for image analysis suggests the need for testing the suitability and accuracy of the algorithms developed. One of the possible approaches is testing these systems with experiments of well-known flow properties. Alternatively, tests can be performed by analysing synthetically generated images. The advantage of the latter approach is that there is no need to set up an experiment and the flow field is known in detail. This paper provides some insights into the relationship between results on both real and synthetic images in a turbulent channel flow. We focus on comparing performances of feature tracking, a novel image analysis technique, particle image velocimetry and particle tracking velocimetry. The three techniques have been used to explore first- and second-order statistics. The results are compared to direct numerical simulations of turbulent flow in a channel (Kim J, Moin P and Moser R 1987 Turbulence in channel flow at low Reynolds number J. Fluid Mech. 177 133–66). Feature tracking performances are rather good, even in its purely translational motion model implementation. No constraints on tracer density have to be introduced. More than 3000 velocity vectors per frame were reconstructed. Resulting accuracy and resolution are always comparable to those achieved by the other techniques.