I have worked with such systems in the lab. Depending on the system used, they might look different, but, basically, one can use various image processing techniques to extract the position and orientation of each marker relative to the camera.
These "fiducial markers," as they are sometimes called, usually have high-contrast borders to distinguish them from most other objects. Inside is a unique symbol or pattern that is asymmetrical enough to enable the camera to distinguish its angle in its plane. I have not looked up the math, but all you need is the initial calibration process and enough processing power to do this in real time.
Please see the following link for some videos that I took of software I built using such a system. You might not have the codec, so you have my apologies ahead of time if this is the case.
Reader Comments (Page 1 of 1)
Vortex @ Dec 23rd 2006 6:59PM
I have worked with such systems in the lab. Depending on the system used, they might look different, but, basically, one can use various image processing techniques to extract the position and orientation of each marker relative to the camera.
These "fiducial markers," as they are sometimes called, usually have high-contrast borders to distinguish them from most other objects. Inside is a unique symbol or pattern that is asymmetrical enough to enable the camera to distinguish its angle in its plane. I have not looked up the math, but all you need is the initial calibration process and enough processing power to do this in real time.
Please see the following link for some videos that I took of software I built using such a system. You might not have the codec, so you have my apologies ahead of time if this is the case.
http://mindlab.msu.edu:16080/lsj/ (The rest of the website it outdated; see http://mindlab.org/ for the current site of the lab.)