Face Morphing Attack Generation and Detection: A State-of-the-Art Review

Abstract

Face morphing attacks pose a significant threat to the security and reliability of biometric identity verification systems, particularly in real-world applications such as passport issuance and border control. In a morphing attack, facial features from two or more individuals are blended into a single image that can deceive face recognition systems, allowing multiple individuals to share a biometric identity. This survey provides a comprehensive overview of the literature on face morphing attack detection (MAD). It begins by introducing the practical implications of morphing attacks and their relevance in operational contexts. The paper then explores a wide range of morphing generation techniques, including both classical landmark-based approaches and modern deep learning-based methods such as GANs and diffusion models. The work proceeds with an extensive review and categorization of existing MAD techniques, grouped into meaningful categories based on their underlying principles, ranging from texture-based and quality-based methods to deep learning and hybrid approaches. By organizing the literature and identifying current trends, strengths, and limitations, this survey offers valuable insight for researchers and practitioners seeking to understand, evaluate, or develop robust solutions against face morphing attacks.