Real time eyeball movement detection based on region division and midpoint position

Abstract

The development of technologies that utilize eye movements as interface for motion detection has an important role in human-computer interaction especially as a medium for controlling automated device such as electrical wheelchair. The reliability of eye movement detection can determine the system performance which can be implemented into an automated device. By tracking the movement of the eyeball, communication between users and automated devices can take easily and may be used by people with hand-foot impairment. In this paper, we propose a method by using the division of the eye region and checking the eyeball midpoint position for detecting the direction of eyeball movement which can be used as a navigation in real time condition We also implement Haar Cascade method for detecting the eye region followed by tracking using Kernelized Correlation Filter to produce stable movement. This approach was proved reliable and could increase the precision and recall on eyeball movement detection significantly, reach the average to 0.93 and 0.84. Even, our proposed method enables a suitable level of sensitivity with low computational time.