A simple calibration method to quantify the effects of head movements on vision-based eye-tracking systems

摘要

A simple method for the calibration of eye-tracking systems based on eye-features detection is proposed. Many eye-tracking systems are claimed to be “robust” to head movements in some degree, but such “robustness” is rarely fully quantified. Moving chin-rest and synthetic images have been proposed to investigate head movements. However, both methods have several limitations. The reported calibration method exploits for the first time ocular imaging eye models (OIEMs) in order to simulate the eye(s), thus allowing to easily investigate the effects due to head movements. Moreover, it can be useful for the development and comparison of eye-tracking hardware, software and whole systems. Differently from human subjects, OIEMs are standardized and time-invariant thus, comparisons can be performed at different times and places, leading OIEMs to be a potential “gold standard test” for eye-tracking systems. Preliminary verification performed on a basic eye tracking system corroborate the applicability of described method.