If I have the intrinsic (camera matrix & distortion coefficients) and extrinsic (rotation + translation) parameters of 2 cameras with respect to a common and fix 3D target (e.g. points in a chessboard).
Is there a way to use bundle adjustment to refine the camera pose and/or the transformation matrix between both cameras? I would definitively prefer a library that avoid me to spend hours understanding the mathematical problem for time issues (e.g. something like the old cv::LevMarqSparse implementation, that avoids me having to know the maths behind the problem).
I know this questions was already made in 2012 here: Bundle adjustment functions
So far I know SBA, SSBA & PBA, but they have not been updated in years.
There was also a OpenCV implementation cv::LevMarqSparse, but it has not documentation and it has been removed in the OpenCV 3.x release.
And so far Ceres Solver is the only option, but it requires a more manual set up and mathematical knowledge of the problem.
Aucun commentaire:
Enregistrer un commentaire