I would like to know a general formula which relates the field-of-view (FoV) to the size of an image and focal length.
The FoV formula for standard cameras is (see e.g. here)
where f is the focal length.
That formula can be derived from the relation which relates the distance between the image point and the principal point, r, and the angle between the principal axis and the incoming ray,
, which reads:
However, the above formula only works for standard cameras with perspective projection. Wide-angle fisheye cameras can have different relations , see for instance here different projections.
The only example in the web I found about FoV for fisheye cameras is in this page, where they say that the FoV is given by
which would be consistent with the above reasoning assuming the equisolid angle projection from this paper.
Is it safe to assume that we can compute the FoV just by inverting the relation for each specific projection, as done with standard perspective projection? Is there any rigorous proof for that?