Track noise reduction

Aim

To reduce the effect of noise in the data during tracking.

example When the center point of an animal is detected in a zone, you want the pellet dispenser to drop a pellet. If the detected center point is moving rapidly because of noise, this may result in a number of consecutive pellets to be dropped, every time the center point crosses the border of the zone. Track noise reduction may solve this problem.

important Do not use Track noise reduction if you are particularly interested in rapid movements of the subject, for example, if you study the startle response of zebrafish larvae.

Background information

If the detected center point of the subject is continuously moving, while in fact the subject is sitting still, the total distance moved will be overestimated. You can use track smoothing to correct for this after you have acquired your data (see Smooth the Tracks), however in some cases you may want to smooth the track during acquisition. This may especially be the case if you use Trial and Hardware Control.

The example below shows the effect of track noise reduction (TNR) on the x coordinates of the tracked subject. The chart plots the x coordinate against time. With TNR on, the path is smoothed out.

Procedure

1.In the Detection Settings, under Advanced, open Smoothing.

2.Next to Track noise reduction, do one of the following:

▪Select Off (default) if you do not want to apply track noise reduction.

▪Select On if you want to apply track noise reduction.

Notes

▪Track noise reduction makes use of the Gaussian Process Regression method. Track noise reduction is applied during acquisition. Hence, it alters the acquired tracks, which cannot be undone afterwards.

With Gaussian Process Regression, the sample points are smoothed, using the x,y coordinates of the previous 12 sample points. This differs from the Lowess post-acquisition smoothing method that uses samples before and after the sample point to be smoothed. This is not possible during acquisition, because the x,y coordinates of future samples are not yet known.

▪Using Track noise reduction in the Detection Settings influences the acquired track, and therefore it is not possible to change it back after acquisition. This is in contrast to post-acquisition smoothing (see Smooth the Tracks) where you can use profiles to calculate analysis results with and without those filters applied.

▪If you use noise-tail tracking, the paths of the nose point and tail base are smoothed independent of the path of the center point.

▪If you use Track noise reduction, the sample points may lag behind when compared with the video images, especially when the subject moves fast. This is because the sample position is smoothed using point locations acquired in the previous samples.

▪When you use Track nise reduction in combination with hidden zones, the subject may go missing when it is in fact in a hidden zone. This occurs when the last (smoothed) position of the subject before it disappears from view is outside the entry zone. See the Troubleshooting topic: The animal is in the hidden zone but the data show “missing sample”.