File:Goniometer principle for camera calibration.jpg

English: Camera calibration is the process of finding the intrinsics of a camera, i. e. the principle point H, the principle distance c, and distortion parameters. In a laboratory this can be done by using a goniometer. At the image plane (focal plane) a plate with an illuminated grid is placed, e. g. instead of the film of an analog camera. The goniometer is placed in front of the lens. At the beginning the goniometer finds the autocollimation point, that is the point where the goniometer sees its own cross hairs which is visible on the illuminated grid. This position is the principle point H. Now the plate inside the image plane ist shifted such, that the cross hairs coincides with one grid point. Then the next grid position is focused and one can measure the angle α with the goniometer. Due to the fact that the grid space d is known one can use the tangens in order to compute the respective angle α´: tan( α´) = c / d. The difference between both angles (Δ = α - α´) is caused by distortion of the lens. But before we have to compute the principle distance c. Another fact is that radial-symmetric distortion is equal where the radius (around H) is constant. So we can use four measurements (left, right, up, and down) in order to compute two unkowns. That gives us an overdetermined equation system. So usually after all measurements are done one performs an adjustment algorithm to compute the principle distance and the parameters of radial-symmetric distortion.