Individual Marker Sorting

[genetunik]

Dear all,
I'm am seriously considering buying the optitrak. I will use it for research purposes to study human motor control (mostly finger-hand-arm motion). I have some reservations about the fact that if I only track individual markers (rather than a rigid body), the marker assignment will be arbitrary from frame to frame. I imagine this must be a common problem and am wondering if anyone has developed a fix, or can suggest an algorithm. I do not need to do this in real time (though it would be nice to have that as well) .... but it is absolutely critical at least for off-line processing.
Any insight is very appreciated.
Best,
Eugene Tunik, PhD, PT
Dept of Rehab and Movement Science
University of Medicine and Dentistry of New Jersey

[Birch]

Tracking markers between frames can be accomplished by looking for the nearest marker in the next frame to the marker's location in the current frame. This tends to produce ok results for slow moving markers which are widely distributed, but has trouble dealing with fast moving markers or very tight clusters. Adding in compensation for marker trajectory can improve the results somewhat.

However, tracking with Rigid Bodies will give much more reliable and robust results

[genetunik]

Thank you for your reply.
1) by "slow moving" do you mean any movements moving below the frequency of the sampling rate?
2) My biggest use of the markers will be to track finger motion....given that sometimes the fingers come together very close, do you think it will be problematic?
3) Do you think that I can fit 3 of the tiny hemisphere markers on each segment of the fingers to define them as rigid bodies?
Thanks,
Eugene

[Birch]

The farther a marker moves per sample, the larger the search area you will need to use to find the marker in the next frame. The larger search area you use, the more likely it is that other markers will be found, leading to potential swaps (especially if there is a high marker density relative to your search area size). So, slow would be relative to both the marker speed, sample rate and marker density.

Using a small rigid body on each finger might work, though it would require a good calibration of the volume and potentially adjusting some of the 3D reconstruction parameters. We plan to be replacing the small hemisphere markers currently in our store with greatly improved facial markers very soon.

[genetunik]

can you please explain what the facial markers will be like and what advantages they will have over the hemispheres?

also, when is soon (1 month, 1 year....)

[Birch]

The old small markers just got removed from our store tonight and replaced with the new ones.

The new facial markers are now smaller, adhesive backed, and should be slightly more hemispherical. They are up for pre-order and should be shipping within the next three weeks (potentially much sooner).

http://www.naturalpoint.com/optitrack/s ... mocap.html