Robust Eye Detection Using Self Quotient image

Abstract

We propose a novel method of eye detection that is robust to obstacles such as the surrounding illumination, hair, glasses and etc. The obstacles above the face images are the constraints to detect eye position. These constraints affect the performance of face application systems such as face recognition, gaze tracking, and video indexing system. To overcome this problem, our method for eye detection consists of three steps. In preprocess, we apply SQI (self quotient image) to the face images to reduce illumination effect. Then, we extract the eye candidates by using the gradient descent which is simple and fast computing method. Finally, the classifier which has trained by using AdaBoost algorithm selects the eyes from all of the eye candidates. The usefulness of proposed method has been demonstrated in experiments with the eye detection performance