Revision history [back]

Note sure if I got it correctly but If you have a stereo camera pair, then why don't you just apply solvePnP for the image captured by the 'second' (say the Right) camera. This way you'll get the R+T for this camera. Note that the 'distCoeffs' may well be different to that of the 1st camera, and obviously it has to be known to use the solvePnP function. In case you want the R+T of the 2nd camera with respect to the 1st camera, then I think you have to multiply the R+T matrix of the 1st camera with the inverse R+T matrix found for the second camera, where R+T matrix is the transformation matrix in homogeneous coordinates (for example, see here, or here). OpenCV has function for affine transformations (look here)

Note sure if I got it correctly but If you have a stereo camera pair, then why don't you just apply solvePnP for the image captured by the 'second' (say the Right) camera. This way you'll get the R+T for this camera. Note that the 'distCoeffs' may well be different to that of the 1st camera, and obviously it has to be known to use the solvePnP function. function.

In case you want the R+T of the 2nd camera with respect to the 1st camera, then I think you have to multiply the R+T matrix of the 1st camera with the inverse R+T matrix found for the second camera, where R+T matrix is the transformation matrix in homogeneous coordinates (for example, see here, or here). OpenCV has function for affine transformations (look here)