FACE AUTHENTICATION METHOD USING EYE TRACKING-BASED HAAR-LIKE FEATURES

摘要

The present invention relates to a face authentication method using haar-like features based on eye tracking. The present invention: recognizes and tracks the face of a person entering a surveillance area; detects a face by applying a haar cascade to a shot image; detects an eye area and pupils in the detected face area; and generates an authentication value by extracting feature points of the face based on pupils. The present invention comprises: (a) a step of generating a haar cascade by learning a face image; (b) a step of receiving an input of a circular image photographed through a camera; (c) a step of detecting a face by applying the circular image to the haar cascade and setting the detected face image in an interest area; (d) a step of detecting an eye area and a nose area in a divided area by equally dividing the interest area; (e) a step of detecting pupils by using a binarization and a hough transform in the eye area; (f) a step of extracting feature points of a face based on the detected pupils; and (g) a step of generating an authentication value by using the extracted feature points to store the same. The face authentication method like the above can authenticate a face quickly and accurately by recognizing and tracking the face of a person entering a surveillance area, detecting a face by generating a haar cascade, and generating an authentication value by detecting pupils.;COPYRIGHT KIPO 2018