An algorithm that identifies meaningful changes or pauses in head movement. Written using Matlab. Data is head rotation recorded using a Vive Pro Eye VR Headset.
Left shows the original rotation data.
Middle shows the potential segments in the data (coloured lines).
Right shows the segments identified by the algorithm.
Segments are identified using a peak analysis on the head velocity at each data point.
However, some peaks are false positives and so need to be filtered out.
One method is to use a simple threshold filter and say that certain characteristics under a threshold indicate a true or false segment.
For example, sharp angles are a strong indication of sudden changes in head movement.
Another method would be to use a probablistic approach, where each characteristic contributes to an overall p-value.
Then if this p-value is significant, the peak can be considered a true positive.
Alternatively, cluster analysis, like k-means can be used to determine true positives.
