Pradeep Kambhampati MEE10:28, pp. 47. ING/School of Engineering, 2010.
The project consists of finding general guidelines for selecting sensors in the application of eye tracking. In eye tracking the sensors delivers images that the system uses. The system finds both the so called glints that have high intensity values and the pupil that has quite low intensity values. This puts high demands on the sensors regarding sensitivity. Other sensor parameters that also affect the actual image quality are also studied and quantified in this project. It has been concluded that the spatial noise and blurring properties mostly affect the eye tracking system’s quality. Different kind of algorithms that finds the sub pixel coordinates of the glints were considered where the moment method gave the best result. An algorithm that calculate the pupil parameters was also presented and compared to a reference system that Tobii use today with about the same performance.