Cameras distort visual reality.
A deterministic model to calculate the distortion would use the characteristics of the lenses, the capturing surface and external distances.
For given photos the characteristics are not known. In such a case it is common to use a first order distortion approximation.
It is common practice to assume spherical distortion so that:
We know the value of " rd " on the images. Unfortunately we do not know the value of the " k " constant or " r " .
Also, we have distortions across two images so that
I suspect that it would be reasonable to assume that k1=k2
We know the geometries of the images, and we have matching points between the images so that it is theoretically possible to determine k.
I have not yet been able to solve this 6th order polynomial analytically (Not even with the help of Sage ). I hope to return to this problem and solve it at a later time.