Estimation of projection errors in human ocular fundus imaging.

Photogrammetric analysis of features in human ocular fundus images is affected by various sources of errors, for example aberrations of the camera and eye optics. Another--usually disregarded--type of distortion arises from projecting the near spherical shape of the fundus onto a planar imaging device. In this paper we quantify such projection errors based on geometrical analysis of the reduced model eye imaged by a pinhole camera. The projection error found near the edge of a 50 degrees fundus image is 24%. In addition, the influence of axial ametropia is investigated for both myopia and hyperopia. The projection errors found in hyperopia are similar to those in emmetropia, but decrease in myopia. Spherical as well as ellipsoidal eye shapes were used in the above calculation and their effect was compared. Our results suggest that the simple spherical eye shape is sufficient for correcting projection distortions within a range of ametropia from -5 to +5 diopters.