TITLE:
Texture Analysis for Makeup-Free Biometrics: A Solution for Imposture Mitigation
AUTHORS:
R. Logeswari Saranya, K. Umamaheswari
KEYWORDS:
Makeup Classification, Makeup Elimination, Convolutional Neural Network, GAN Models, Haar-Cascade Algorithm
JOURNAL NAME:
Open Access Library Journal,
Vol.12 No.2,
February
28,
2025
ABSTRACT: Face recognition is rapidly becoming one of the most popular biometric authentication methods. Most face recognition systems are focused on extracting features and enhancing their verification and identification capabilities. The detection of security vulnerabilities of different types of attacks has been given attention only in recent years. These attacks can include, but are not limited to: Obfuscation Spoofing and morphing; for example, a hacker can masquerade as a target to gain access to the biometric system. The application of cosmetics can alter the appearance of a face, leading to a decreased characteristic distinctiveness. Facial makeup includes variations in skin tones, the position of eyebrows and skin complexion. The cosmetic effect on an individual causes the face recognition system to falsely identify the person affecting the security of the biometric system. Adding a presentation attack detection module to the existing biometric system can be the solution to this problem. In this work, a CNN-based machine learning approach is adapted to classify the presentation attack using texture analysis. The proposed method is to extract the original face by removing makeup so that the FR system recognizes the person’s real identity, resulting in decreased vulnerability. The false accept rate (FAR) is a measure of a biometric system’s resistance to zero-effort attacks and is generally considered as the system’s performance.Subject AreasArtificial Intelligence, Computer Vision