Risk of Fingerprints Online Exposure: Importance of Spoofing Detection and Prevention

Resumo

Introduction: The use of fingerprints as a biometric method has been growing. This growth is due to their uniqueness, immutability, and ease of use in many contexts. Fingerprints are widely used as an authentication method for bank accounts, cell phones, doors, among others, being tempting to steal and spoof for illicit acts, with the aggravating factor that once a fingerprint is compromised, we cannot change it like we change a code or password (Uliyan et al., 2020). The increased risk of images exposed online from which fingerprints can be extracted (Jara San Miguel et al., 2021) is a concerning factor.

Aim: Analyse the scientific literature on fingerprint spoofing and prevention to identify limitations and research gaps that require further investigation.

Materials and Methods: A search was conducted in three scientific databases: PubMed, IEEE Xplorer and Wiley, to find articles on fingerprint spoofing, spoofing techniques and materials, and anti-spoofing software. To achieve this goal, terms like “fingerprint” AND “spoof*” AND “personal recognition” were used. Articles without peer review, editorials, reviews and duplicates and with more than 7 years old were excluded.

Results: Normally, common materials are used to create the spoof, like Play-Doh, silicon, glue, wax, among others. However, it is also possible to use complex polymers that need to be created (Saguy et al., 2021). In most cases, the fake fingerprints were able to deceive traditional sensors; however, when these sensors were accompanied by deep learning software or liveness detectors, such as blood flow, skin distortion or odor detectors, among others, the success rate was generally much lower (Uliyan et al., 2020; Lu et al., 2020).

Discussion: Once the perpetrators gain access to fingerprints, creating a spoof is straightforward, typically using common materials to make a mould and produce the final spoof (Kanich et al., 2018). The fact that simple spoofs are able to pass sensors like the ones we use to protect our phones raises an important question: what can we do to protect ourselves? In response to this question, some studies developed deep learning software to aid in the sensor signal processing (Uliyan et al., 2020; Lu et al., 2020). It greatly decreased the success rate of the spoofs; however, a new problem appeared: it only works properly when the spoof is made of a material that the machine was trained to detect, making it a matter of guessing ahead of time possible spoofing materials (Uliyan et al., 2020).

Conclusion: The increasing use of fingerprints as a biometric authenticator highlights the growing importance of addressing spoofing threats. Current anti-spoofing methods are effective primarily against materials included in training datasets, underscoring the need for further research to improve detection across a wider range of spoofing materials.