NEC says biometric ‘fluctuation’ estimation preserves privacy and accuracy

Matching face biometrics with consistent accuracy but without stored facial information has long posed a challenge for hashing systems, but NEC says it has found a way. The veteran biometrics company has come up with a digital signature technology that utilizes biometric information to generate and authenticate cryptographic keys, ensuring accuracy by correcting for differences in how the facial image is captured.

The company says this achieves a “safer and more secure authentication” by reducing the risk of facial information leakage. Further, the technology can be applied not only to face biometrics but to all types of biometric authentication.

The technology relies on generating the same personal key from fluctuating face biometrics during registration and authentication. This is because the facial biometric information captured during authentication can differ from the time of registration — due to factors such as the angle of the face, face expression, and shooting environment, which can lead to the risk of generating a different key despite it being the same individual.

NEC developed a unique algorithm that generates the individual’s key even from facial information that changes with each capture. The algorithm estimates the degree of fluctuation in facial information at the time of capture compared to the registered data. If the estimated fluctuation is significant, it will determine it as another person and prevent the generation of another person’s key. NEC claims this allows for “high-precision authentication” compared with traditional face biometric recognition, and enables fast processing capable of handling large-scale authentication.

Furthermore, NEC developed a “secret computing method” that allows verification while keeping the generated keys protected. This method enables the entire authentication process to be executed in an encrypted state, the company says, and so key leakage is eliminated. The company provides some detail in an R&D post on its website (in Japanese).

The technology can generate digital signatures compliant with standard digital signature methods such as ECDSA and EdDSA, making it usable for proving the authenticity of electronic documents, contracts, Trade of digital assets, and presentation of digital certificates, NEC added.

ZeroBiometrics introduced its facial authentication without biometric template storage last year.

NEC will present practical applications of the new technology first in payments then in entry/exit applications this year, with an introduction to the technology at Retail Tech Japan 2025 (March 4-7) in Tokyo.

Article Topics

biometric authentication  |  data privacy  |  face biometrics  |  facial authentication  |  facial recognition  |  NEC