Reputation: 69
I have two cameras with different resolutions but stereoCalibrate functions has only one option for imagesize.
If I understand correctly stereoCalibrate computes rigid transformation matrix from cam1 to cam2. If that is true then which camera size should I use as input to the function stereoCalibrate?
Upvotes: 5
Views: 3612
Reputation: 5354
The imageSize
parameter is used only to initialize intrinsic camera matrices.
I suggest to calibrate each camera independently using cv::calibrateCamera()
and so get the camera matrices and distortion coefficients for each camera. And then estimate the transformation between the camera coordinate systems (rotation R
and translation t
) using cv::stereoCalibrate()
with flags CV_CALIB_FIX_INTRINSIC
enabled (with the pre-estimated camera matrices and distortion coefficients).
And so the imageSize
parameter doesn't matter anymore.
Upvotes: 4