Method for Measuring Rotational Eye Movement by Tracking Eye Vascular Images Using luminance Gradient Information

Abstract

To estimate the angle of the rotational eye movement, the use of the vascular image of the white of the eye has an advantage of being less influenced by a change in its ambient environment unlike the use of iris patterns. However, it has a disadvantage of being hard to select any of eye vascular images as a template suitable for accurate tracking for the reasons; 1) a comparatively strong light source in its ambient environment is reflected on the eye surface and 2) the shape of the vessel in the white of the eye considerably vary with individuals. Accordingly, the present study proposes a method for estimating the rotational eye movement using a template image, which is obtained through the process involving the steps of; extracting image feature points from the white area of the eye in a captured image based on a luminance gradient and selecting, as a template image, the area of the image having a lot of feature points with less reflection from its ambient environment and less erroneous estimation because of no similar shape of vascular image in a search area. The result of the evaluation of the method showed the average value and standard deviation of all the errors of estimation were -0.69° and ±1.02°, respectively in various lighting environments.