Iris Style Transfer: Enhancing Iris Recognition with Style Features and Privacy Preservation through Neural Style Transfer

TL;DR

This work introduces iris style transfer by extracting iris style features via neural style transfer (NST) and applying NST to obfuscate identifiable iris patterns for privacy preservation. The authors demonstrate that iris style features, represented by global per-channel statistics, outperform traditional CNN embeddings in recognition and exhibit greater resilience to rotation and perspective distortions. They further show that stylizing iris textures substantially reduces recognition accuracy for both feature types while preserving eye segmentation and gaze estimation utilities, and they quantify a manageable increase in false acceptance risk. The study provides a concrete privacy-utility trade-off framework and discusses practical extensions toward real-time deployment and broader biometric privacy applications.

Abstract

Iris texture is widely regarded as a gold standard biometric modality for authentication and identification. The demand for robust iris recognition methods, coupled with growing security and privacy concerns regarding iris attacks, has escalated recently. Inspired by neural style transfer, an advanced technique that leverages neural networks to separate content and style features, we hypothesize that iris texture's style features provide a reliable foundation for recognition and are more resilient to variations like rotation and perspective shifts than traditional approaches. Our experimental results support this hypothesis, showing a significantly higher classification accuracy compared to conventional features. Further, we propose using neural style transfer to obfuscate the identifiable iris style features, ensuring the protection of sensitive biometric information while maintaining the utility of eye images for tasks like eye segmentation and gaze estimation. This work opens new avenues for iris-oriented, secure, and privacy-aware biometric systems.

Figures (12)

• Figure 1: Two examples of neural style transfer. Eye image sources from OpenEDS2019 garbin2019openeds.
• Figure 2: The iris style feature extraction pipeline. Eye image sources from OpenEDS2019 garbin2019openeds.
• Figure 3: Style reconstructions using a single layer each time. The raw iris sources from OpenEDS2019 garbin2019openeds dataset.
• Figure 4: Test performance of style feature and common CNN feature for iris recognition.
• Figure 5: Examples of image variations. The raw iris sources from OpenEDS2019 garbin2019openeds dataset.