Theoretical and Applied Cybersecurity (TACS)

Постійне посилання на фонд

Зараз показуємо 1 - 1 з 1

Відкритий доступ
Development of an anti-spoofing method for images in biometric security systems using ML
(НН ФТІ, КПІ ім. Ігоря Сікорського, 2025) Palahin, V.; Palahina, O.; Ivchenko, O.; Protsenko, O.; Bairak, A.
The integrity and technical security of biometric authentication systems are critical in applications that require protection against mobile access to high-security areas. The main technical security issue is the vulnerability to spoofing attacks, where fraudulent biometrics deceive the system. This study proposes to use the ratio of facial landmarks from realtime video streams to improve live person detection in security systems. By analyzing responses to interactive prompts and calculating facial landmark ratios, the system aims to more reliably distinguish between genuine and fake interactions. This approach leverages advanced real-time image processing techniques to enhance accuracy in environments where traditional static biometric systems often fail. Utilizing dynamic response analysis further helps in mitigating the risks associated with spoofing by requiring active participation from the subject, therefore adding an additional layer of security.