Hierarchical Perceptual Noise Injection for Social Media Fingerprint Privacy Protection

Billions of people share images from their daily lives on social media every day. However, their biometric information (e.g., fingerprints) could be easily stolen from these images. The threat of fingerprint leakage from social media has created a strong desire to anonymize shared images while maintaining image quality, since fingerprints act as a lifelong individual biometric password. To guard the fingerprint leakage, adversarial attack that involves adding imperceptible perturbations to fingerprint images have emerged as a feasible solution. However, existing works of this kind are either weak in black-box transferability or cause the images to have an unnatural appearance. Motivated by the visual perception hierarchy (i.e., high-level perception exploits model-shared semantics that transfer well across models while low-level perception extracts primitive stimuli that result in high visual sensitivity when a suspicious stimulus is provided), we propose FingerSafe, a hierarchical perceptual protective noise injection framework to address the above mentioned problems. For black-box transferability, we inject protective noises into the fingerprint orientation field to perturb the model-shared high-level semantics (i.e., fingerprint ridges). Considering visual naturalness, we suppress the low-level local contrast stimulus by regularizing the response of the Lateral Geniculate Nucleus. Our proposed FingerSafe is the first to provide feasible fingerprint protection in both digital (up to 94.12%) and realistic scenarios (Twitter and Facebook, up to 68.75%). Our code can be found at https://github.com/nlsde-safety-team/FingerSafe.

I. Introduction

Posting photos on social media is a popular way to share our daily lives with others. However, given that thousands of publicly shared images on Instagram contain accessible fingerprint details [1], personal biometric information (e.g., fingerprints, etc) can be easily stolen from photos shared on social media, which may cause severe security problems for fingerprint authentication systems (e.g., access control system) as shown in Fig. 1. There is extensive evidence to support the feasibility of the above challenges. For example, hackers easily accessed the fingerprint of the president of the EU Commission via an online photo in 2014;1 recently, it was reported by CNN and BBC that fingerprints have been easily extracted from images shared on social media,2, 3 even from images taken from 3 meters away.4 The fingerprint leakage is irreversible—since you cannot change your fingerprint, once the fingerprint has been leaked, all systems that rely on fingerprints are at risk for the rest of your life [2]. Based on these leaked fingerprint images, hackers can gain authorized assess to the access control systems of governments, banks and the police,5 or payment systems such as ApplePay6 with more than an 80% success rate with an inkjet-printed paper [3] or a 3D printed mold [4]. The security of fingerprint-based systems is currently at severe risk. ¹

https://www.theguardian.com/technology/2014/dec/30/hacker-fakes-german-ministers-fingerprints-using-photos-of-her-hands

²

https://www.bbc.com/news/uk-wales-43711477

³

https://edition.cnn.com/2021/05/25/uk/drug-dealer-cheese-sentenced-scli-gbr-intl/index.html

⁴

https://www.ft.com/content/446ac29a-dbc1-11e6-9d7c-be108f1c1dce

⁵

https://www.forbes.com/sites/zakdoffman/2019/08/14/new-data-breach-has-exposed-millions-of-fingerprint-and-facial-recognition-records-report/?sh=38315c146c60

⁶

https://arstechnica.com/information-technology/2020/04/attackers-can-bypass-fingerprint-authentication-with-an-80-success-rate/

Fig. 1.

Posts on Facebook unconsciously leak fingerprint of owners. The protection of pixelization is limited, large PGD is effective but unnatural. In contrast, FingerSafe is effective and natural. Images used with consent from owners.