A Real-Time Lazy Eye Correction Method for Low Cost Webcams

Abstract

Lazy eye, or Amblyopia, affects 3.2% of the adult population, Amblyopia can reduce the clarity of vision within the affected eye of the sufferer, causing sufferers to obtain side-effects such as a drift, in the weaker eye, from the point of focus. Eye-tracking systems are becoming more prevalent within topic areas such as human-computer interaction and action assistance such as road sign detection. The accuracies of eye-tracking systems must be ensured to be of highest caliber. Therefore, the impact of Amblyopia upon the accuracies of eye-tracking must be considered and discussed. This paper discusses the effects of Amblyopia on the accuracy of a recent eye tracking system and presents a real-time correction method, which implements the analysis of collected calibration data from the user, using nearest neighbor calibration points for calculation of predicted drift at the current focus of a user. The experiments we conducted on participants suffering from Amblyopia show promising results of our method.