VideoTrack
2D
Motion capture by marker tracking
What
is VideoTrack2D?
VideoTrack 2D is a motion capture software package for Windows. It tracks
retroreflective markers placed on the analysed body and outputs their
2D coordinates.
How are markers detected in the video?
Through arbitrary chains of image processing filters. For example each
video frame can be filtered with a color filter, then blurred to remove
noise and finally thresholded. After processing the image only contains
white marker pixels and black background. Marker pixel cluster-centers
determine marker positions.
How
does marker identification work? What about occlusions or ghost markers?
Starting from an initial manual marker assignment, each marker is tracked
in the video sequence. Marker occlusions or bright spots in the video
(ghost markers) can disturb the tracking process, as well as fast marker
movements with non smooth trajectories. VideoTrack contains several tracking
algorithms to improve tracking in these situations:
- Next neighbour tracking: Movement of the marker is determined by its
next neighbour in the following frame.
- Permutation tracking: All possible marker identifications are evaluated
in terms of minimal displacement, minimal distance changes between them
and minimal angular changes between lines connecting markers.
- Hybrid tracking: Each marker is tracked by one of previous algorithms
What about manual re-assignment?
VideoTrack 2D supports two editing modes for manual marker identification.
Also particular markers can be exluded from analysis at frame intervals
or set to constant positions.
How does deinterlacing work?
PAL video frames are saved in interlaced video format, which means 25Hz
time resolution for a full video frame or 50Hz divided into odd and even
video frames. VideoTrack's deinterlacing algorithm splits each frame in
even and odd lines and linearly interpolates missing lines. It also supports
batch deinterlacing for more than one file at a time. |