Dear Community,
usually one wants to stitch images together to create a (spherical) panorama. However I want to do the opposite for a project of mine. That is, subsampling images from a panorama image. Let's say we have a continous (spherical) 180° image. I would like to subsample images from any angle (yaw, pitch, roll) within the possible range of the spherical image. But how am I supposed to know which pixel-length corresponds to what angle?
Is there an implementation of opencv to do that or can someone point me to the math behind this?
The panorama was created by me using the stitcher class of opencv in python. I'm kind of stuck as the stitcher returns no information on the panorama, such as the sphere radius it was warped to. Is there also a way to get more information on the stitching process with the python api?
Cheers, D