Nowadays, the importance of the so called smart devices is growing significantly, extending the functionality of the original product in order to make the everyday life more convenient. The revolution of the smart devices started by the enhancement of the traditional cell phones by extra functions, examples include WiFi connection, GPS and voice recognition system. The tendency was not uniquely characteristic for mobile phones, several different smart devices gained ground, as smart watches or glasses. Another important branch of these devices was developed for vehicles, especially voice recognition systems, integrated navigation systems or fatigue detection.
Extending the functionality of the aforementioned devices, the so called gaze tracking systems are growing more and more popular. These are capable of extracting the direction of the gaze, or in certain cases the location of the examined object can be calculated based on the observation of the subject’s eye. One of the leading applications is the hands-free display, which not only serves as a convenient interface, but also able to help disabled users in everyday life. Furthermore, gaze trackers can be utilized to check the fatigue level of the driver, or provide useful information about the surroundings and the traffic.
Most of the gaze trackers use the pupil’s position to extract the gaze point. The accurate detection of the pupil is crucial in this method, as a minute deviation in the pupil’s position may mean a substantial difference in the gazing distance. The main goal of the thesis is to improve the pupil center detection algorithm of an existing gaze tracker in order to achieve higher accuracy.