I tried similar methods like optimize camera intrinsics by sfm described in paper, but incorrect camera intrinsics will lead incrroect camera pose and landmarks. It is hard to make intrinsic parameters converge.
do we need a close approximate of the real value?
I tried similar methods like optimize camera intrinsics by sfm described in paper, but incorrect camera intrinsics will lead incrroect camera pose and landmarks. It is hard to make intrinsic parameters converge.
do we need a close approximate of the real value?