WiFaKey: Generating Cryptographic Keys from Face in the Wild

TL;DR

WiFaKey tackles the challenge of deriving cryptographic keys from face biometrics in unconstrained environments by combining AdaMTrans, an adaptive random masking feature transformation, with Neural-MS, a learning-based LDPC decoder. The system converts real-valued face features into binary templates, aligns their error characteristics with LDPC capabilities, and robustly recovers keys in open-set, noisy conditions. Extensive experiments across six unconstrained and two constrained datasets show competitive Genuine Match Rates at zero False Match Rate, with strong unlinkability and quantified security bounds via leftover entropy. The work demonstrates practical biometric key generation improvements and establishes a foundation for secure, privacy-preserving face-based cryptographic keys in real-world deployments.

Abstract

Deriving a unique cryptographic key from biometric measurements is a challenging task due to the existing noise gap between the biometric measurements and error correction coding. Additionally, privacy and security concerns arise as biometric measurements are inherently linked to the user. Biocryptosystems represent a key branch of solutions aimed at addressing these issues. However, many existing bio-cryptosystems rely on handcrafted feature extractors and error correction codes (ECC), often leading to performance degradation. To address these challenges and improve the reliability of biometric measurements, we propose a novel biometric cryptosystem named WiFaKey, for generating cryptographic keys from face in unconstrained settings. Speciffcally, WiFaKey ffrst introduces an adaptive random masking-driven feature transformation pipeline, AdaMTrans. AdaMTrans effectively quantizes and binarizes realvalued features and incorporates an adaptive random masking scheme to align the bit error rate with error correction requirements, thereby mitigating the noise gap. Besides, WiFaKey incorporates a supervised learning-based neural decoding scheme called Neural-MS decoder, which delivers a more robust error correction performance with less iteration than non-learning decoders, thereby alleviating the performance degradation. We evaluated WiFaKey using widely adopted face feature extractors on six large unconstrained and two constrained datasets. On the LFW dataset, WiFaKey achieved an average Genuine Match Rate of 85.45% and 85.20% at a 0% False Match Rate for MagFace and AdaFace features, respectively. Our comprehensive comparative analysis shows a signiffcant performance improvement of WiFaKey. The source code of our work is available at github.com/xingbod/WiFaKey.

Figures (14)

• Figure 1: Overview of the WiFaKey. Our proposed feature transformation pipeline, i.e., the AdaMTrans, transforms features generated from extractors into binary templates. The commitment is generated and stored in the database; a neural min-sum decoder (the lower block) is trained and serves as the decoder to retrieve the key from the de-committed codewords.
• Figure 2: Diagram of the AdaMTrans.
• Figure 3: Frame error rate vs. crossover rate and number of iterations.
• Figure 4: Ablation study on $\tau$ concerning accuracy on transformed binary features (CALFW).
• Figure 5: Ablation study on $\tau$ with respect to key retrieval accuracy (CALFW, 100 iterations).