A new implementation method of ASEF for eye detection

Eye detection is one of the most important steps in the face-related systems such as the face recognition, gaze tracking, and eye blink detection. The various methods for eye detection have been proposed for a long time. In this paper, we proposed the new implementation method for ASEF to detect eye positions precisely. The normal face images are used to produce the ASEF, and eye positions are detected by rotating the ASEF by up to -10, -5, +5, +10 degrees. To validate the proposed method, the experiment is conducted on our c (ETRI). It has not only frontal face images, but also slightly rotated face images. Also it contains the various hairstyles, glasses, facial expressions and illumination. In the experimental result, the detection rate of our method has been improved about 10% than the original implementation method.