Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

MorCode: Face Morphing Attack Generation using Generative Codebooks

Aravinda Reddy PN, Raghavendra Ramachandra, Sushma Venkatesh, Krothapalli Sreenivasa Rao, Pabitra Mitra, Rakesh Krishna

TL;DR

This paper tackles the vulnerability of face recognition systems to morphing attacks by introducing MorCode, a 2D morphing generator conditioned on a VQGAN codebook. It blends latent representations of two faces via spherical interpolation and decodes the result to produce high-quality morphs, supported by a new MorCode Morphing Dataset (MMD) built from FRGC V2 data. Comprehensive experiments against three deep FRS (ArcFace, MagFace, AdaFace) show MorCode achieves the highest attack potential across digital and print-scan data, underscoring the strength of codebook-conditioned morphing. The work provides open-source code and a dataset to facilitate reproducibility and the development of effective defenses against morphing attacks in biometric security pipelines.

Abstract

Face recognition systems (FRS) can be compromised by face morphing attacks, which blend textural and geometric information from multiple facial images. The rapid evolution of generative AI, especially Generative Adversarial Networks (GAN) or Diffusion models, where encoded images are interpolated to generate high-quality face morphing images. In this work, we present a novel method for the automatic face morphing generation method \textit{MorCode}, which leverages a contemporary encoder-decoder architecture conditioned on codebook learning to generate high-quality morphing images. Extensive experiments were performed on the newly constructed morphing dataset using five state-of-the-art morphing generation techniques using both digital and print-scan data. The attack potential of the proposed morphing generation technique, \textit{MorCode}, was benchmarked using three different face recognition systems. The obtained results indicate the highest attack potential of the proposed \textit{MorCode} when compared with five state-of-the-art morphing generation methods on both digital and print scan data.

MorCode: Face Morphing Attack Generation using Generative Codebooks

TL;DR

This paper tackles the vulnerability of face recognition systems to morphing attacks by introducing MorCode, a 2D morphing generator conditioned on a VQGAN codebook. It blends latent representations of two faces via spherical interpolation and decodes the result to produce high-quality morphs, supported by a new MorCode Morphing Dataset (MMD) built from FRGC V2 data. Comprehensive experiments against three deep FRS (ArcFace, MagFace, AdaFace) show MorCode achieves the highest attack potential across digital and print-scan data, underscoring the strength of codebook-conditioned morphing. The work provides open-source code and a dataset to facilitate reproducibility and the development of effective defenses against morphing attacks in biometric security pipelines.

Abstract

Face recognition systems (FRS) can be compromised by face morphing attacks, which blend textural and geometric information from multiple facial images. The rapid evolution of generative AI, especially Generative Adversarial Networks (GAN) or Diffusion models, where encoded images are interpolated to generate high-quality face morphing images. In this work, we present a novel method for the automatic face morphing generation method \textit{MorCode}, which leverages a contemporary encoder-decoder architecture conditioned on codebook learning to generate high-quality morphing images. Extensive experiments were performed on the newly constructed morphing dataset using five state-of-the-art morphing generation techniques using both digital and print-scan data. The attack potential of the proposed morphing generation technique, \textit{MorCode}, was benchmarked using three different face recognition systems. The obtained results indicate the highest attack potential of the proposed \textit{MorCode} when compared with five state-of-the-art morphing generation methods on both digital and print scan data.

Paper Structure

This paper contains 9 sections, 6 equations, 3 figures, 4 tables.

Table of Contents

  1. Introduction
  2. MorCode: Proposed 2D Face morphing generation
  3. Encoder
  4. Spherical Interpolation
  5. Vector Quantization
  6. Decoder
  7. MorCode Morphing Dataset (MMD)
  8. Experiments and Results
  9. Conclusion

Figures (3)

  • Figure 1: Block Diagram of the proposed method MorCode to construct high quality morphed images.
  • Figure 2: Example images from MMD dataset representing Digital samples. The proposed MorCode face morphing technique is qualitatively compared with the five different existing techniques.
  • Figure 3: Example images from MMD dataset representing print scan using DNP printer samples. The proposed MorCode face morphing technique is qualitatively compared with the five different existing techniques.