Automatic identification of eye movements using the largest lyapunov exponent

Abstract

Abstract The study of eye movements has been increasing over the past decade. It is considered that eye movements, mainly saccades and blinks, provide significant information for cognitive and visual processes of the observers. Saccades and blinks are high velocity eye movements. In this paper, the automatic identification of saccades and blinks, as well as their onset and offset, is proposed based on a novel implementation of nonlinear dynamics using the Largest Lyapunov Exponent and the logarithm of the divergence. The Largest Lyapunov Exponent detection method was tested on 25,000 saccades and 2,366 blinks, detecting with high accuracy and precision both types of eye movements. The Largest Lyapunov Exponent detection method was compared against two other existing techniques for blink and saccade identification, showing advantageous performance.