Author(s): Rui M. Lage Ferreira; Luis Mendes; Ana Margarida Ricardo; Alexandre Bernardino

Linked Author(s): Ana Margarida Ricardo

Keywords: Optical Flow; PIV; Fluid Mechanics

Abstract: Optical Flow (OpF) approaches are quite diverse. In Fluid Mechanics, OpF methods are frequently applied to images that have been collected with laser or led illumination and seeded flows (n short images acquired for Particle Image Velocimetry). In these cases, it is important to know which OpF methods are more adequate for which combinations of flow types, imaging conditions and image characteristics. This paper addresses this practical need by assessing the performance of three OpF methods, Lucas and Kanade (1981), Horn and Schunck (1981) and Farneback (2003), combined with the Liu and Shen (2008) algorithm. We evaluate the accuracy of different OpF methods as the difference between their values and of ground-truth. Hydrodynamic conditions include deformation dominated, rotation-dominated and uniform flows. For each flow type, relative and absolute errors are computed for different tracer displacements, noise powers, pixel particle sizes, image bit-depths and particle concentrations. The accuracy of the OpF methods is mainly affected by the magnitude of velocity gradients and convective accelerations. The inner region of the Poiseuille flow and the saddle point in the Rankine vortex combined with uniform flow pose significant difficulties to all methods. The accuracy of the Lucas-Kanade/Liu-Shen combination is high for all flow types, image conditions image bit depths. The Farneback/Liu-Shen combination has a similar high performance but only for images of 10 bit or higher. The Horn-Schunck algorithm is the worst performing method, due to high sensitiveness to particle concentration variations or particle sizes.

DOI: https://doi.org/10.3850/IAHR-39WC2521711920221558

Year: 2022