Doing a camera calibration of an extremely distorted camera ( wide-angle ) the OpenCV gives us the rogdriguez and the transpose vectors of each image used in the calibration process. How do they do this?
Assuming I have a distorted image of a planar target with blobs on it I want to find the target orientation.
In undistorted case it is easy to determine its orientation by using the cv2.findHomography function.
One idea I had was to take a center 4 blobs close to the middle of the image hoping that the distortion in the middle of the camera was not pronounced as it is further away but I think this did not yield a precise homogropahy similar to the one that was found by the OpenCV.