DSMOOTH - Filtering video-tracked animal paths in orientation research

Post-processes animal paths by removing tracking gaps and pauses, smoothing and downsampling.

dsmooth is a Matlab function for post-processing of video-tracked animal paths, mainly for use in orientation research. The main purpose is to remove tracking errors and small animal movements that are of no importance to the studied behaviour. To achieve this, the function first tries to remove tracking gaps and pauses in the animal track. The path is then smoothed to reduce tracking errors. Finally, the path is downsampled to a minimum segment length between consecutive points of the path, in order to make the paths (and in particular their lengths / tortuosities) comparable across animals moving at different speeds, animals of different species, and different tracking frequencies. The function was originally developed to study the orientation behaviour of ball-rolling dung beetles (Smolka et al., 2016).