Evaluation of Particle Tracking Codes for Dispersing Particles in Porous Media

Particle tracking (PT) is a common technique in microscopy, microfluidics, and colloidal transport studies, where time lapse images are recorded at a high frame rate and image analysis is used to reconstruct particle trajectories. The performance of many PT codes has been tested for particles that exhibit Brownian-type motion. However, PT is frequently used to track particles in porous media where heterogeneous velocity fields generate non-Brownian motion of particles.

To understand the capabilities and limitations of PT algorithms in heterogeneous flows, we simulated the advective transport of particles in an open channel and in two confined channels that differ in their packing and grain size distribution. We tested four different PT codes that differ in their feature finding, trajectory linking and trajectory filtering algorithms. For each geometry, we performed a sensitivity analysis with respect to particle density, particle speed and particle intermittency. We used the ground truth trajectories and imagery from each simulation to compare the performance of the different PT codes by calculating the average Euclidean distance between ground truth and tracked trajectories, the probability distributions of trajectory lengths and velocities, and the false positive rate for spot detection. We show that increasing particle density, speed, and intermittency will negatively impact all PT codes, but some codes are consistently impacted less. We also find that our statistics reveal different problems with trajectory linking for each PT code. Furthermore, our statistics can be used to determine the best PT method for advected particles in a particular porous geometry.