Eye centre localization using low cost web cam

Abstract

Numerous Computer vision applications uses eye tracking and face recognition. An eye tracker is far better input then other input devices. According to the historical point of view first approach on camera based eye tracking was illustrated by Dodge and Cline in 1901. Eye tracker is all based on the analysis of person’s eye movement, attention, feelings and concentration. The main objective is to achieve eye tracking with more precision using low cost web camera. But recent systems which depend on available light to predict eye centre using various algorithms fail to detect eye centre in low resolution or low cost. To achieve this objective software which is used is opencv3.0 and visual studio 2015 with low cost and low resolution camera. To achieve this objective Image gradient is one of the best approaches used to predict eye centre accurately. By following this simple approach function in this paper which has only dot product makes possible to achieve the objective. The position of the maximum of this function is at the point where maximum gradient vectors intersect and thus to the eye’s centre. This method helps to detects eye centre localisation of pupil in 99% of the cases in spite of varying contrast pose or illumination.