Facial De-morphing: Extracting Component Faces from a Single Morph

Banerjee, Sudipta; Jaiswal, Prateek; Ross, Arun

doi:10.1109/ijcb54206.2022.10007977

Cited by 10 publications

(9 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The restoration accuracy on the HNU‐FM dataset is presented in Table 4, and some examples are in Figures 7 and 8. It can be seen that DAD [32] achieved a lower recovery accuracy because it only used morphed images. But its latent capacity should be valued.…”

Section: Resultsmentioning

confidence: 99%

“…Perhaps it is because it cannot choose the appropriate de‐morphing factors. And [20, 32] do not achieve the high restoration effect. Although FD‐GAN [19] achieves a relatively high similarity in the recovered faces, there are issues such as local blurriness in the images.…”

Section: Resultsmentioning

confidence: 99%

“…Ref. [32] proposed a blind de‐morphing method that can recover two real facial images from a single given morphed facial image without reference images or prior knowledge. Recently, a detection and localization method for face MAs is proposed based on feature‐wise supervision by [16], and designed a fine‐grained classification loss to supervise the network, enhancing its robustness against unknown MAs.…”

Section: Related Workmentioning

confidence: 99%

“…However, the above-mentioned methods focus on detecting whether an image has been morphed and do not discuss how to restore the original facial images of accomplices. Recent studies [18][19][20]32] have started to explore the restoration of the accomplice's facial image. For instance, [18] proposed a method called face de-morphing, which is the inverse process of face morphing.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

ADFF: Adaptive de‐morphing factor framework for restoring accomplice's facial image

Long,

Zhou,

Zhang

et al. 2023

IET Image Processing

View full text Add to dashboard Cite

Morphing attacks (MAs) pose a substantial security threat to the Automatic Border Control (ABC) system. While a few morphing attack detection (MAD) methods have been proposed, the face morphing accomplice's facial restoration has not received sufficient attention. Due to the inability to foresee the morphing factor used for a particular morphed image, selecting the appropriate de‐morphing factor becomes a challenging problem in the restoration of the accomplice's facial image. If the morphing factor cannot be chosen reasonably, achieving the desired restoration effect is difficult. Therefore, this paper presents an adaptive de‐morphing factor framework (ADFF) architecture for restoring the accomplice's facial image. By exploiting the morphed images stored in the electronic passport system and the real‐time captured criminal's images, ADFF can effectively restore the accomplice's facial image. Experimental results and analysis show that ADFF can significantly reduce the security threats of MAs on ABC.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Resultsmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

ADFF: Adaptive de‐morphing factor framework for restoring accomplice's facial image

Long,

Zhou,

Zhang

et al. 2023

IET Image Processing

View full text Add to dashboard Cite

show abstract

“…In [35], a conditional GAN is designed to disentangle identity from the morphed image using the pixel difference by minimizing conditional entropy. Recently, [36] proposed a method to recover both bona fide face images simultaneously from a single given morphed face image without reference image or prior knowledge. Such blind demorphing is conceptually similar to the unmixing of hyperspectral images.…”

Section: De-morphingmentioning

confidence: 99%

MorphGANFormer: Transformer-based Face Morphing and De-Morphing

Zhang¹,

Li²,

Li³

et al. 2023

Preprint

View full text Add to dashboard Cite

Semantic face image manipulation has received increasing attention in recent years. StyleGAN-based approaches to face morphing are among the leading techniques; however, they often suffer from noticeable blurring and artifacts as a result of the uniform attention in the latent feature space. In this paper, we propose to develop a transformer-based alternative to face morphing and demonstrate its superiority to StyleGANbased methods. Our contributions are threefold. First, inspired by GANformer, we introduce a bipartite structure to exploit long-range interactions in face images for iterative propagation of information from latent variables to salient facial features. Special loss functions are designed to support the optimization of face morphing. Second, we extend the study of transformer-based face morphing to demorphing by presenting an effective defense strategy with access to a reference image using the same generator of MorphGANFormer. Such demorphing is conceptually similar to unmixing of hyperspectral images but operates in the latent (instead of pixel) space. Third, for the first time, we address a fundamental issue of vulnerability-detectability trade-off for face morphing studies. It is argued that neither doppelganger nor random pair selection is optimal, and a Lagrangian multiplierbased approach should be used to achieve an improved trade-off between recognition vulnerability and attack detectability.

show abstract