Fast eye localization based on a new haar-like feature

This paper focuses on fast eye localization method. According to the priori proportional relationships of face features, we firstly set an appropriate candidate window from the face region detected. Secondly, histogram equalization is applied on the candidate region to eliminate illumination effects. This paper presents a new haar-like feature generating the confidence of the feature throughout the candidate region in order to locate eyeball accurately and rapidly. Our method is proved to be simple and robust against the disturbance caused by glasses, eyebrow and hair. The process of training and learning is not necessary because of the appropriate priori knowledge. Our experiment on three face databases shows that our method can be applied to real time eye localization and even to pupil localization under most circumstances, achieving accurate results.