International Journal for Advanced Research in Science & Technology

Face recognition systems have become an essential component in secure real-time applications such as automated border control (ABC) systems.However, face morphing attacks are becoming a challenge to these systems because a morphed image, generated by morphing the facial features of two or more individuals, can bypass biometric authentication.Such attacks pose a serious security threat because they enable unauthorized people to obtain electronic machine-readable travel documents (eMRTDs) or epassports, which allow them to pass through borders without detection.This paper provides an in-depth review of face morphing attacks, considering both the generation and detection of morphed images.Several morph generation techniques, including image warping, deep learning models, and hybrid approaches, are discussed against state-of-the-art methods for morph attack detection, that include forensic analysis, deep learning-based detection, and cross-domain techniques.Benchmarking efforts and public databases are also reviewed against evaluation metrics for MAD systems.The paper discusses the current challenge of realism in the generated morphs, robust detection over diverse conditions, and considerations of privacy.Finally, potential future directions such as real-time detection, adversarial training, and multimodal biometric approaches are explored in order to address these emerging threats for biometric security systems.