Adaptive fuzzy commitment scheme based on iris-code error analysis

Biometric cryptosystems is a group of emerging technologies that securely bind a digital key to a biometric so that no biometric image or template is stored. Focusing on iris biometrics several approaches have been proposed to bind keys to binary iris-codes where the majority of these approaches are based on the so-called fuzzy commitment scheme. In this work we present a new approach to constructing iris-based fuzzy commitment schemes. Based on intra-class error analysis iris-codes are rearranged in a way that error correction capacities are exploited more effectively. Experimental results demonstrate the worthiness of our approach.