Difference between cv.homographyBasedEstimator.apply() and cv.findHomography()

Hi OpenCV community,

Could I ask what is the difference between cv.homographyBasedEstimator.apply() and cv.findHomography()? I find that findHomography produces a ‘better’ matrix to stitch images together (despite differences in rotation and scale). I am trying to convert the stitching_detail.py file into one that accomodates findHomography instead, however findHomography does not have a pairwise matching utility iirc.

it’s not supposed to. use the feature matching parts of OpenCV for feature matching. that means FLANN in general, and the bruteforce matcher if you’ve just read a trash “tutorial”.

the stitching module is newer than findHomography. it does a lot of things in its own way, because it has to.

these two things aren’t interchangeable. each has its purpose.