0

I'd like to know if my understanding is correct regarding camera focal length.

The intrinsics of the camera consists the focal length (and of course other parameters, not focusing on them here..) which is really derived from the two parameters Fx and Fy, where in cameras that provide us square images the parameters of the focal length hold the following attribute: Fx=Fy=F (for my understanding, not 100% sure regarding this), unlike rectangular images case, where Fx and Fy differ.

Is it correct that this occurs from the fact that the width and height of an object will be reduced in different proportions according to the distance between the object and the camera?

If my claim does hold, than if i am to calibrate a camera where the image i'm receiving has very big width/height ratio, than i should anticipate that Fx and Fy will differ largely?

Thanks in advance.

Nir Yakobovits
  • 11
  • 1
    This seems completely mixed up. I don't know where you got this idea about $F_x$ and $F_y.$ Can you point to a source that people can examine to see what it really says (or indeed if the source itself isn't wrong)? You can have a lens with astigmatism so that it has two different focal lengths, but that lens will be unable to give you a sharp picture on any shape of film at any distance. – David K Dec 11 '21 at 19:24
  • Hey thanks for the comment, and the source came from here: https://www.codetd.com/en/article/10862432 Nonetheless, could you elaborate more regarding these parameters? Why isn't the focal length is just F, and the reasoning for dividing it to Fx and Fy? – Nir Yakobovits Dec 12 '21 at 07:26
  • The linked page is seems a little confused itself; for example, a pinhole camera theoretically has no focal length. (You can place the film at any distance from the pinhole.) But it seems we are dealing with a particular matrix in a particular software package where the symbols $f_x$ an $f_y$ have little to do with a plane of focus but instead provide a linear transformation of the image. This is specifically an OpenCV question, not a question about math or even about a typical actual camera. – David K Dec 12 '21 at 15:48
  • The closest analogy I can think of in actual cameras is an anamorphic lens. When wide-screen movies were distributed on actual film, the camera would take a scene with a width much greater than its height and fit it onto a frame of film that was more nearly square (4:3 ratio). Then in projection they would expand the width more than the height. If you look directly at the image on the film, however, it would look squashed from side to side. – David K Dec 12 '21 at 15:55
  • I think when the web page speaks of a "square" image it has nothing to do with the number of pixels between the boundaries of the image in either direction; it has to do with whether if there is a sphere in the scene, the image of the sphere has the same number of pixels of height and width. – David K Dec 12 '21 at 15:57

0 Answers0