python - Homography? OpenCV (cv2) method to handle missing chessboard corners in some views? -
i have installed opencv 3.2.0 , can find many/most of corners in this image this question playing around contrast using cv2.findchessboardcorners(left_gray, (6,5))
or using cv2.goodfeaturestotrack(gray_img,25,0.01,10)
described in the answer.
while findchessboardcorners either returns 6×5=30 of them or none
, can see lower quality images not points going found with the goodfeaturestotrack
method.
but when read documentation cv2.calibratecamera()
looks object points need correspond on one-to-one basis image points each , every view. if 1 of images has missing point, i'd have find , remove object point list image.
is there cv2 method automatically handle this?
edit: i'm wondering, cv2.findhomography()
using 1 of robust methods i'm looking here?
of course if external row or column missing, there unresolvable ambiguity, if few internal points missing, calibration should still possible.
i try write something, , of course reject view wrong number of found corners, if there existing way handle non-ambiguity-causing missing points, i'd try first. goal automate process as possible without skipping every image wrong number of found points.
Comments
Post a Comment