Measurement of three dimensional eye position using image processing: a geometric approach

Accurate measurement of 3-dimensional eye position is an important tool in the study of the human oculomotor system. The expense and invasive nature of existing 3-dimensional measurement systems have prompted the development of video-based systems for eye position measurement. Most video systems use some variation of the polar cross correlation method, but this technique produces inaccurate results for the torsional component of eye position if the spherical shape of the eye is not taken into account. We have extended the polar cross correlation technique by considering eye geometry and developing the formulae required to determine the projection of the eye into the image plane for all eye positions. These formulae also allow the representation of 3-dimensional eye position in Fick-angles or as rotation vectors, which are commonly used in oculomotor research.<
>