Leveraging Diffusion For Strong and High Quality Face Morphing Attacks

TL;DR

The paper tackles the security risk posed by face morphing attacks on modern FR systems. It introduces a diffusion-based morphing pipeline that jointly exploits semantic and stochastic latent representations, optimized with a DDIM scheduler, to produce high-fidelity morphed faces containing dual-identity information. Through extensive experiments on three FR systems and three datasets, the diffusion approach demonstrates superior visual fidelity (low FID) and stronger vulnerability to FR verification, while remaining challenging for existing MAD systems; a novel relative-strength metric is proposed to compare attacks. These results underscore the need for more robust MAD methods and inform future directions toward higher-resolution morphing and diffusion-enhanced defenses.

Abstract

Face morphing attacks seek to deceive a Face Recognition (FR) system by presenting a morphed image consisting of the biometric qualities from two different identities with the aim of triggering a false acceptance with one of the two identities, thereby presenting a significant threat to biometric systems. The success of a morphing attack is dependent on the ability of the morphed image to represent the biometric characteristics of both identities that were used to create the image. We present a novel morphing attack that uses a Diffusion-based architecture to improve the visual fidelity of the image and the ability of the morphing attack to represent characteristics from both identities. We demonstrate the effectiveness of the proposed attack by evaluating its visual fidelity via the Frechet Inception Distance (FID). Also, extensive experiments are conducted to measure the vulnerability of FR systems to the proposed attack. The ability of a morphing attack detector to detect the proposed attack is measured and compared against two state-of-the-art GAN-based morphing attacks along with two Landmark-based attacks. Additionally, a novel metric to measure the relative strength between different morphing attacks is introduced and evaluated.

Figures (7)

• Figure 1: Example of the proposed Diffusion-based morphing attack. Samples are from FRLL dataset.
• Figure 2: The forward and reverse Diffusion processes.
• Figure 3: Proposed architecture for Diffusion-based morphs, where the green traces indicate variables associated with identity $a$, likewise red traces denote identity $b$, and blue traces for the morphed identity $ab$.
• Figure 4: Comparison across different morphing algorithms of two identity pairs from the FRLL dataset.
• Figure 5: Comparison of Diffusion and MIPGAN-II morphed faces on FRLL. Images are resized to $256\times256$ and cropped to $224\times224$ to match VGGFace2 pre-processing pipeline.