Registration of human eye movement in biometric research

One of the basic human sense is vision. Every day most of the people use their eyes for gathering information about surrounding world. The light is constantly being captured by the human eye, sampled, causing the eye to move to sample next important light structure. In next step the brain is processing the information provided by the eye. Scientists all over the world try to make similar process artificial, allowing computers to gain from sense of vision. Great number of research has been put into eye tracking making it more and more efficient over the years. In this paper we try to propose a solution which is in position to track eyes. There are used two USB cameras connected to a Desktop PC. Each camera tracks one eye and basing on position of center of the eye compute its coordinates. In biometric projects quality of the input signal is crucial variable for whole process. In research was checked 3 algorithms: color filtering, circle Hough transform and optical flow. Each of approached which was tried has its advantages and disadvantages. It would be difficult to have good eye tracking system basing only on one algorithm. So in the work it was proposed a method based on combination of optical flow for tracking the center of the eye, while resetting its position was obtained using Hough transform every Xth frame. It eliminated problem of drift and blinking as was shown on numerical examples.