pyPTV - a complete framework to reconstruct physical flow fields from camera images

Abstract

We introduce a novel probability-based Particle Tracking Velocimetry (PTV) and Data Assimilation (DA) framework called pyPTV. It is an open-source project under continuous development and is completely written in the programming language Python. The framework provides a complete routine for processing measurement data suitable for PTV. Starting from raw camera images, pyPTV includes image processing, calibration, and a novel PTV algorithm to reconstruct individual seeding particle trajectories (tracks) in a Lagrangian frame of view. In addition, post-processing algorithms are implemented to refine the tracks. The framework includes interpolation and DA algorithms, that allow the user to generate physical flow fields in an Eulerian frame of view. The velocity field of the particle tracks is corrected to be solenoidal, and the pressure field is estimated. The implemented debugging scripts help fix problems with each subroutine during processing. To verify the method, a synthetic test case of turbulent Rayleigh-Bénard (RB) convection (Pr = 6.9, Ra = 1E10) in a cubic cell is generated by direct numerical simulation with massless tracer particles. A reconstruction accuracy of over 65% with 95% correct tracks is achieved at tracer particle densities of about 0.1 ppp