I am looking for a way to approach a computer vision problem I'm having.
I have working tracking system:
- 4-8 cameras
- Gives (x,y,z) of a infrared led
- Each led Transmits a unique 8 bit signal
The tracking system is expensive and the interface is too hard for our users to work with. I want to replace it with a possible my own/ OpenCV implementation.
My current approach which seems to require a lot of development of what seems to be common problems:
- Calibrate the cameras to make a 3D space - The cameras need to know where they are in space and in relation to each other.
- Given two or more camera sees a unique led it uses gray-scale image with the pixel to calculate the 3D position (x, y, z) of that led.
Right now I am attempting to write my own custom algorithm for both task and its proving to be a lot of work. Is it possible to approach this with OpenCV to help with the heavy lifting.
Take a look at Free track : http://www.free-track.net/english/ you can download sources there.
来源：https://stackoverflow.com/questions/18493487/infrared-led-tracking-using-opencv-to-track-x-y-z-positions