A New Pupil Detection Algorithm Based on Multiple Angular Integral Projection Functions

In this paper a new pupil detection algorithm based on multiple angular integral projection functions is presented. The algorithm is based on binarizing the eye pupil image captured with an infrared video camera, using the dark pupil technique. To better detect the features points that belong to the pupil, the proposed algorithm uses the angular integral projection function, applied along six directions. The paper analyzes the cases when the six directions are evenly distributed around the center of the pupil in fixed positions, as well as when they are rotated with a random offset. These cases are analyzed and discussed in the paper, presenting the results obtained on three representative databases of eye images.