Towards Chaos-driven Max-Min Hashing for Palmprint Template Protection

Template protection mechanisms enhance the security deployment capabilities of palmprint recognition systems by applying irreversible transformations or encryption techniques to biometric templates. However, most existing methods rely on fixed auxiliary data or transformation parameters to maintain consistency between registration templates and query templates, introducing a self-contradictory vulnerability: once these external elements are compromised, irreversibility is lost. To address this issue, we propose a user-centric, chaos-based palmprint template protection framework named CHAMP, which eliminates the need for any auxiliary storage or parameter sharing. CHAMP integrates two key components: Chaos-Driven Temporary Encrypted Data (CDT-ED) , which generates chaos-based permutations and matrix groups during the session and immediately discards them, enforcing a short-lived “use-once-and-destroy” strategy to prevent the reuse or reconstruction of conversion traces; Chaotic Maximum-Minimum Index Hashing (CMMIH) , an algorithm combining chaotic feature disruption with maximum-minimum index encoding to blur spatial structures and enhance irreversibility; Experimental results on fingerprint datasets from PolyU, Tongji University, and IITD demonstrate that CHAMP achieves state-of-the-art recognition performance. Additionally, comprehensive security analysis confirms its compliance with revocable biometric standards, providing robust irreversibility, unlinkability, revocability, and privacy protection against various attacks.