track_RAFT

Fast accurate particle tracking even for large particle displacements

track_RAFT

Fast accurate particle tracking even for large particle displacements

Many physics applications require the tracking of multiple objects between images. Examples could be tracking individual bacteria swimming around under the microscope, or tracking individual birds in a swarm. For this sort of thing, the processing flow would be to first locate the individuals in each frame, and then perform tracking between frames to connect trajectories.

The standard go-to tracking code was developed by Crocker and Grier, and converted into Matlab by Blair and Dufresne (track.m). This code works well when individual objects do not move very far between frames (relative to the average interparticle distance). However, it cannot cope with large displacements, and gets extremely slow when there are large numbers of objects per frame.

This code is designed to work with large displacements, by tracking cluster patterns of particles. It is designed to have the same syntax as track.m, so it can be replaced easily in standard codes. It works best when there are a reasonable number of objects in every frame - do not use it when you have < 20. As far as I have experienced, there are far fewer false tracks that crop up with this code, and it is significantly faster than track.m

Examples of the code's use can be found in [1,2]. For tracking large displacements in solids, we have found it useful to combine this code with a step to remove obviously wrong tracks, with the function remove_outliers_RAFT.

[1] Kim et al. PRX 2021 (https://arxiv.org/abs/2103.04975)

[2] Testa et al. Nature Commun. 2021