Abstract

This research establishes a system that can detect eye fatigue in the early stages, which leads to a non-invasive simulation of Electrooculography (EOG) signals. To generate physiologically realistic EOG signals for normal and eye-fatigue situations, a physiological EOG model with baseline drift, Gaussian blink pulses, and controlled noise is utilized. The artificial signals are detrended and band-pass filtered to get rid of the artifacts and make the blinks visible. The optimized threshold and distance parameters are employed to find the blink peaks. The principal features such as blink rate, amplitude, duration, RMS, and variance are used for the eyes' classification as normal or fatigued. Actually, in the blink rate, amplitude, duration, RMS, and variance changes of the fatigued signals, which correspond to the physiological fatigue patterns that have been cited in the literature, can be found in all cases. The results demonstrate that simulated EOG signals can be utilized to represent the characteristics of fatigue and thus can be regarded as a reliable source for early eye fatigue detection systems.