DCMorph: Face Morphing via Dual-Stream Cross-Attention Diffusion

2026-04-23

AI Summary

Existing face fusion attack methods often struggle to emulate realistic threats due to fusion artifacts or reconstruction distortions. This work proposes a dual-stream diffusion fusion framework that, for the first time, enables explicit dual-identity conditioning within diffusion models and introduces a geometry-consistent initial latent representation. By jointly operating on identity conditions and the latent space, the method integrates decoupled cross-attention interpolation, DDIM inversion, and spherical latent interpolation to significantly enhance fusion fidelity and attack stealth. Experimental results demonstrate that the approach achieves the highest attack success rates across four state-of-the-art face recognition systems and effectively evades current detection mechanisms.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Search and Optimization: Mixed Discrete/Continuous Search

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search

Abstract

Advancing face morphing attack techniques is crucial to anticipate evolving threats and develop robust defensive mechanisms for identity verification systems. This work introduces DCMorph, a dual-stream diffusion-based morphing framework that simultaneously operates at both identity conditioning and latent space levels. Unlike image-level methods suffering from blending artifacts or GAN-based approaches with limited reconstruction fidelity, DCMorph leverages identity-conditioned latent diffusion models through two mechanisms: (1) decoupled cross-attention interpolation that injects identity-specific features from both source faces into the denoising process, enabling explicit dual-identity conditioning absent in existing diffusion-based methods, and (2) DDIM inversion with spherical interpolation between inverted latent representations from both source faces, providing geometrically consistent initial latent representation that preserves structural attributes. Vulnerability analyses across four state-of-the-art face recognition systems demonstrate that DCMorph achieves the highest attack success rates compared to existing methods at both operational thresholds, while remaining challenging to detect by current morphing attack detection solutions.

Problem

• face morphing
• morphing attack
• identity verification
• diffusion models
• face recognition

Innovation

• dual-stream diffusion
• cross-attention interpolation
• identity-conditioned latent diffusion
• DDIM inversion
• face morphing attack