Resetting your fingerprints

The obvious problem with biometrics is that once someone has stolen your fingerprint or iris ID, you cannot simply reset those to block their access as you might a password. Now, research in the International Journal of Computational Vision and Robotics offers a new approach to protecting biometric authentication data so that the risk associated with this kind of irreversible identity theft can be largely avoided and give users an option to reset their fingerprints and other biometrics, as it were.

Biometric authentication systems identify individuals using physiological or behavioural characteristics, such as fingerprints, facial features, or even how they type or move a computer mouse rather than passwords or physical tokens. However, such traits are essentially fixed and so compromised data cannot simply be reset like a password. To address this, the study focuses on the idea of cancellable biometrics, a technique in which biometric data is deliberately transformed from the start so that it can be revoked and replaced if stolen, while still allowing accurate identity verification.

The proposed system combines several computational techniques to protect biometric templates. Feature extraction is performed using Speeded-Up Robust Features (SURF), a computer vision method that detects distinctive points in images. The resulting data is then processed using a Fast Fourier Transform (FFT), a mathematical tool that converts signals into frequency components. Security is further enhanced through index-of-maximum hashing, which encodes dominant features into compact representations, and a matrix-based operation used to combine vectors securely.

The team has tested their approach on standard fingerprint datasets and found it to be comparable with existing methods but stronger than some in terms of strengthening resistance to attacks, including record multiplicity attacks, where adversaries attempt to reconstruct original data by linking multiple compromised templates.

Shaikh, A.S. and Patel, V.D. (2026) ‘Fingerprint template protection: cancellable biometrics’, Int. J. Computational Vision and Robotics, Vol. 16, No. 4, pp.399–410.