Optimized Transformation and Matching for Iriscode Based Cancelable Biometrics Systems

Iris recognition is a reliable biometric model for identity identification, but reliability issues and cross-matching concerns can affect its accuracy. This research proposes a secure and cancelable iris template protection scheme to address these issues while maintaining system performance, privacy, and security. The scheme includes a matching technique for speed optimization and a transformation technique using locality sensitive hashing (LSH). The proposed framework improves the variance of original hamming distance in Iriscodes, discriminates between real and fake matches, and reduces vertical dependency, achieving an EER of 0.46% in empirical tests on CASIA v3-internal iris databases.