Multi-Camera Calibration for Motion Tracking

I am trying to calibrate two cameras using Matlab 2014a and the example found here: http://www.mathworks.com/help/vision/ref/extrinsics.html This is for use in a low-cost motion tracking system. From the calibration, I want to ultimately obtain each camera's absolute position in the world coordinate system.

In order to get position, I did the following: Convert transposeMatrix from a 1x3 to 3x1 position = -transpose(rotationMatrix)*transposeMatrix Is this correct?

Should the calculated positions of the cameras be relative to the coordinate system created in the image I use for "imOrig"?

Thank you for your help.

Answer 1

[Dima Lisin]

I think you meant "translationVector" instead of "transposeMatrix". Is that right?

The rotationMatrix and translationVector give you the transfromation from the checkerboard's coordinate system into the camera's coordinate system. So to find the location of the camera in the checkerboard's coordinate system you have to do the inverse transformation:

position = -translationVector * rotationMatrix';

This will give you the position as a 1-by-3 vector. Also, please keep in mind that all the matrices use the post-multiply convention, i. e. row vector times a matrix.