Trustworthy Compression? Impact of AI-based Codecs on Biometrics for Law Enforcement

TL;DR

This work evaluates AI-based image codecs for compressing biometric data used in law enforcement, across iris, fingerprints, and soft-biometrics. By comparing GAN-based and autoregressive AI codecs (Hific, Mbt, Ms) against JPEG baselines, and by assessing image quality with SSIM as well as recognition performance, it reveals that perceptual fidelity does not guarantee preserved biometric information. Iris recognition degrades notably under AI compression, while fingerprint recognition remains relatively robust; soft-biometrics show texture-dependent degradation. The findings emphasize the need for task-specific compression choices and standards when deploying AI-based codecs in security-sensitive contexts.

Abstract

Image-based biometrics can aid law enforcement in various aspects, for example in iris, fingerprint and soft-biometric recognition. A critical precondition for recognition is the availability of sufficient biometric information in images. It is visually apparent that strong JPEG compression removes such details. However, latest AI-based image compression seemingly preserves many image details even for very strong compression factors. Yet, these perceived details are not necessarily grounded in measurements, which raises the question whether these images can still be used for biometric recognition. In this work, we investigate how AI compression impacts iris, fingerprint and soft-biometric (fabrics and tattoo) images. We also investigate the recognition performance for iris and fingerprint images after AI compression. It turns out that iris recognition can be strongly affected, while fingerprint recognition is quite robust. The loss of detail is qualitatively best seen in fabrics and tattoos images. Overall, our results show that AI-compression still permits many biometric tasks, but attention to strong compression factors in sensitive tasks is advisable.

Figures (8)

• Figure 1: AI codecs can impact typical biometric data for law enforcement depending on the use case. Best viewed in its digital version.
• Figure 2: An iris example from the CASIA-Iris-Thousand dataset CASIAIris compressed at the lowest quality level using different AI Codecs and JPEG.
• Figure 3: Recall for iris recognition on AI-based codecs and JPEG.
• Figure 4: Average Hamming distance from two iris codes of the same original and compressed image.
• Figure 5: Distribution of non-matches and matches depending on the distance from iris recognition for an AI codec with different quality levels.