A feature and appearance based method for eye detection on gray intensity face images

This paper presents a robust and precise eye detection algorithm on gray intensity face images. Our method combines the strength of two existing methods which are a feature based method and an appearance based method to detect and locate a precise pupil center. It includes the following three steps. First, the feature based method is used. The method uses a projection function to detect all possible regions of each left and right eye. Second, the appearance based method is used to filter out all non eye regions, and keep only one region for each left and right eye. All possible regions will go through a principal component analysis (PCA). Third, a modified hybrid projection function is used to locate the pupil center for both eyes. The experimental results show that the proposed method has an efficient performance with the accuracy of more than 90.52%.