A METHOD FOR DETECTING AND DETERMINING FACE FOR IMAGE-BASED SECURITY

摘要

A method of detecting and discriminating a face for image-based security is provided to detect a normal face stably as well as increase reliability by accurately discriminating the wearing of sunglasses or a mask. A method of detecting and discriminating a face for image-based ATM(Automated Teller Machine) security comprises the steps of: sensing the approach of a user according to infrared sensing(S110); extracting a first facial candidate region by using a spatiotemporal model or a background model(S120); extracting a second facial candidate region by using an RPCA(Robust Principal Component Analysis)-based background model and a time-variable image for the extracted first facial candidate region(S130); extracting an facial contour by using an ASM(Active Shape Model) for the extracted second facial candidate region(S140); discriminating eye and mouth regions by using histogram indexing from the extracted facial contour(S150); discriminating the wearing of accessories along the discriminated eye and mouth regions(S160).