Masking agent: researchers develop anti-spoofing method

摘要

Facial makeup can alter a person's appearance, but can it also degrade the accuracy of automated face recognition? Arun Ross, a researcher at Michigan State University, says he and other researchers have designed a method to automatically detect the presence of makeup in face images, and account for it during matching. The proposed algorithm extracts a feature vector that captures the shape, texture, and colour characteristics of the input face, and employs a classifier to determine the presence or absence of makeup. Alterations due to facial cosmetics are predominately colour-based, he said, but affect the performance of face-matching software based on greyscale images.