2
$\begingroup$

I have a series of about 50 images, each separated by a small rotation and taken from the same camera. The images are feature-dense, and I can calculate a homography from image-to-image easily.

My question is, is there any way to determine the rotation matrix and/or intrinsic camera calibration parameters given just these images?

$\endgroup$
2
$\begingroup$

If the images are related by pure rotation, then the homography would give you the rotation matrix. If the rotation is in-plane, then you would not be able to compute the intrinsics.

$\endgroup$
  • $\begingroup$ Yeah, I was just doing the math and got that for a pure rotation, the homography would be equal to KRK^-1. So that solves that problem. What do you mean when you say the rotation is "in-plane"? $\endgroup$ – James Conlin Mar 9 '15 at 16:11
  • $\begingroup$ In-plane means rotation about the camera's optical axis, aka roll. Out-of-plane rotation involved pitch and yaw. $\endgroup$ – Dima Mar 9 '15 at 16:16
  • $\begingroup$ Ah, got it. All of the rotation is pitch and yaw. Can you give any suggestion how I would got about determining the intrinsic parameters? $\endgroup$ – James Conlin Mar 9 '15 at 16:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.