high accuracy small volume tracking
Posted: Tue Mar 19, 2013 9:31 am
Hi,
I have 12 small squares (53mm x 53mm x 2mm) that I want to track uniquely in a very small volume (see attached figure of setup). Therefore, I attached a unique pattern of small infrared reflective small (flat) infrared reflective markers on top of each tile. On every tile I putted 5 small markers (I used a puncher to get very small, round and same sized markers out of the reflective tape that I have).
My setup consists of 8 flex13 optitrack camera's (see attached figure of setup). During calibration I got exceptional results in medium and high quality mode so I applied the high quality results. However, I needed to turn down the illumination of all the camera's to 5 instead of 15 because false positive markers were detected everywhere (most of them due to IR reflection on the table or hands/arms when users are interacting with the tracked squares).
However, the trackables are still tracked incorrectly many times. Actually, the tracking jitters a lot (e.g. trackables detected in the wrong orientation,...). I set the max marker deflection to 2 and the min hit count to 3 and flexibility to 10. Also setting the residual lower (from 5.x to 1.0) helped a little bit). But still the results are not optimal.
The trackables still jitter and the latency is relatively high because a lot of data is streamed (I think this is because the cameras are really close and thereby the markers are relatively big).
[img:center]https://dl.dropbox.com/u/5276376/setup.png[/img]
Can somebody provide me some tips on how to improve this setup...? Maybe I need to put the camera's further away from each other?
I have 12 small squares (53mm x 53mm x 2mm) that I want to track uniquely in a very small volume (see attached figure of setup). Therefore, I attached a unique pattern of small infrared reflective small (flat) infrared reflective markers on top of each tile. On every tile I putted 5 small markers (I used a puncher to get very small, round and same sized markers out of the reflective tape that I have).
My setup consists of 8 flex13 optitrack camera's (see attached figure of setup). During calibration I got exceptional results in medium and high quality mode so I applied the high quality results. However, I needed to turn down the illumination of all the camera's to 5 instead of 15 because false positive markers were detected everywhere (most of them due to IR reflection on the table or hands/arms when users are interacting with the tracked squares).
However, the trackables are still tracked incorrectly many times. Actually, the tracking jitters a lot (e.g. trackables detected in the wrong orientation,...). I set the max marker deflection to 2 and the min hit count to 3 and flexibility to 10. Also setting the residual lower (from 5.x to 1.0) helped a little bit). But still the results are not optimal.
The trackables still jitter and the latency is relatively high because a lot of data is streamed (I think this is because the cameras are really close and thereby the markers are relatively big).
[img:center]https://dl.dropbox.com/u/5276376/setup.png[/img]
Can somebody provide me some tips on how to improve this setup...? Maybe I need to put the camera's further away from each other?