I am trying to decompose a homography matrix. I read that there are a few methods to do that (e.g. Faugeras, Zhang). But I wasn't able to produce a useful output yet.
I am not sure which form my homography matrix needs for decomposition. A few questions about that:
I read somewhere that the matrix needs to be centered. It needs to be the way so that I can put in a centered picture (center of the picture has coordinates $[0,0]$) and get the new coordinates. Is this true?
Which form of coordinates should the matrix take? $1 ...$ to e.g. $1000$ or $0 ... 1$
Do I have to adjust the matrix before I can use it for example with camera parameters? How?
Hopefully someone has experience with that.
Thanks.
