A High-security-level Iris Cryptosystem Based on Fuzzy Commitment and Soft Reliability Extraction

In this paper, an error-correction-based iris recognition (EC-IR) scheme that guarantees both secure template storage and high-level recognition accuracy is first constructed for personal authentication. From the analysis of both the soft reliability values for the iris bits and the recovery capability values for the low-density parity-check (LDPC) code bits, a method called template mapping is devised in order to freely adjust the error-correction capability of the EC-IR scheme. The design process for suitable LDPC codes is then investigated so as to provide a stable and high-rate EC-IR scheme. In order to further enhance the security level of the EC-IR scheme, we propose locating the dominating feature points (DFPs), and then using them for iris recognition rather than using the original binary templates acquired from the iris database. The DFP-based EC-IR scheme not only enhances the security level, but also provides a better equal error rate (EER) performance and a faster processing speed during the verification stage. As a result, an iris cryptosystem based on the fuzzy commitment strategy is eventually constructed founded on a comprehensive consideration of both a satisfactory recognition performance combined with a high security level.