Deriving stereo extrinsic parameters

[KevinCain]

Thanks for your intriguing work, continuing in the footsteps of Loop & Zhang.

Could you provide a suitable conversion for calibrated stereo camera data to the left+right extrinsic matrices you require as input to example.py -- for example, using OpenCV's cv2.stereoCalibrate output?

[decadenza]

Hi, and thanks for your interest.

OpenCV, for convetion, sets the world origin in the first camera and then you only have the extrinsic matrices (a rotation R and a translation T) that lead you from the 1st to the second camera (see StereoCalibrate documentation).

To solve your problem, for the first camera, you simply have to use: R1 = 3x3 identity matrix t1 = zero array which joined together become:

RT1 = np.array([[ 1, 0, 0,  0],   # Left extrinsic parameters
                    [0, 1, 0, 0],
                    [0, 0, 1, 0]])

And for the second camera, you can use the stereo calibrate output values of R and T, again, joined together in the 3x4 matrix:

RT2 = np.hstack((R, T.reshape(3,1) ))

The opposite can be done too, moving the world origin in the first camera. I'll be soon releasing a stereo vision library containing several tools for these operations and more.

Hope this helps.

[decadenza]

Obviously intrinsic matrices (called cameraMatrix1 and cameraMatrix2 in OpenCV) and distortion coefficients (distCoeffs1 and distCoeffs2) are unaffected by the transformation and remain exactly the same ones!

[decadenza]

I'm going to close this. Feel free to reopen for any issue.

[decadenza]

@KevinCain I published the full library with many examples. For general stereo vision operations, please refer to it SimpleStereo. Examples of calibration are provided too.

Hope this helps many other people working with this stuff.

Cheers