Single-Marker vs. Multi-Markers Traking

Mar 28, 2010 at 9:22 PM


I would like to know if multi-markers tracking is more accurate than single marker tracking.

Does multi-markers tracking take some sort of average from the poses of the different markers or does it just compute the transformation from the first marker detected?

Thank You in advance.

Mar 29, 2010 at 11:25 PM

From my experience, multi-marker tracks better than single marker, and also it's more reliable since even if part of the marker becomes invisible, multi-marker situation still tracks it as long as at least one marker is visible. 

For the details, you should ask VTT people who developed ALVAR directly since I don't get to see their source code.


Mar 30, 2010 at 1:29 AM

Thanks Ohan.

I'm trying to develop an application where a marker is rendered on a LCD screen along with 2D shapes drawn in various positions around the marker.

When I point my camera to the screen, the marker should be recognized and 3D shapes should appear on top of the corresponding 2D ones (ex: a cube on top of a square) by using different transformations for each shape.

However, I noticed that most of the time the 3D shapes are not correctly aligned on top of the 2D ones.

Therefore, I was wondering whether the problem is caused from tracking the marker on screen or it's  just the accuracy of marker tracking in ALVAR?

Do you have any suggestions to achieve better marker tracking and shape positioning  (to have the 3D shapes and 2D align perfectly) in such a scenario? 

Again, thanks a lot for your help.

Mar 31, 2010 at 5:49 PM

It's hard to give a suggestion without seeing how the tracking behaves in your scenario. 

Also, if you have questions regarding the optimal placement of the markers for best tracking, VTT will more likely give you better answer than I do, so please ask them directly.