Self-calibration of zooming cameras observing an unknown planar structures

ICPR 2000
Malis and Cipolla
나도 planar sturucture를 사용해야 할 것 같아서 보다가 토할 뻔함.

어쨌든 2월 12일 몫.

주 내용

it is based on the estimation of the collineations existing between several views of a plane.

even if the structure of the plane is unknown it must be the same for all the images and this provides some constraints which allow the recovering of the varying focal length.

the only prior geometric knowledge on the features that will be used here is their coplanarity. two views of a plane are related by a collineation. using multiple views of a plane we obtain a set of collineations which are not independent.

homography and collineation are generally used to indicate the same projective transformation from Pn to Pn (n=2). we will use the term 'homography' to indicate a collineation expressed in the Euclidean space.

...

the super-collineation matrix...

the super-homography matrix...

each independent homography will provide us two constraints on the parameters.

the total number of constraints which can be obtained from m images is 2(m-1)...

ground truth(by Faugeras-Toscani)를 확보하기 위해 calibration grid 이미지를 사용하였다 (10장).

principal point를 image center로 가정하고, focal lengths에 대해서만 푸는 것 같다.

zoom 없을 경우 +-1% 정도의 오차를, zooming할 경우 _+- 5% 정도의 오차를 보인다.

역시 Zhang을 다시 한 번 보는 게 나을 것 같다... ㅜㅜ