I am trying to utilize stereoRectify on my own image frames collected from a left and right camera that were NOT calibrated with stereoCalibrate. Basically, I have all of the calibrated, matrix information needed for stereoRectify, but the guides I have seen across multiple forums only show how to obtain them with stereoCalibrate, to then be passed into stereoRectify. This is not what I want to do however; I would like to pass in hard-coded matrices myself into stereoRectify. What format should these matrices take in Python’s OpenCV, and how can I exactly go about doing this?

you can find an example of stereo intrinsic params here

(careful with the dist coeffs, they’re in weird `[r1, r2, p1, r3, p2]`

order !)

you’ll also need the `R`

and `T`

matrices (pose between cams). if you have those as a Vec, use Rodrgues to convert to a 3x3 Mat

I think all I have, of the params that you have mentioned/linked, is the R matrix and dist coeffs, but I am not completely sure.

Here is the matrices outputted by the ROS bag file, and the .yaml file that includes the distortion coefficients. Wouldn’t the info from both of these files be sufficient enough to fill stereoRectify, or am I still missing some matrices?

please, if we need to see it, put it AS TEXT, HERE

(not as an image on some external bin… thank you)

@berak Apologies, basically for both cameras I have the distortion parameters (k1, k2, t1, t2, k3), intrinsic camera matrix for the raw images, rectification matrix, projection matrix, and distortion coefficients.

Would that be enough for stereoRectify?

R,T (pose between your cams) is still needed