Pattern-Driven Privacy: Columnar Binary Transformation for Robust Iris Recognition

Iris is a distinct and strong biometric feature that is being employed as a method of authentication all over the world. Irreversibility, Renewability, and Unlinkability are the most critical aspects required to ensure the privacy of biometrics. Keeping recognition performance at a high level while providing privacy is a critical challenge that many researchers try to overcome. Information distortion through transforming original iris data into another form of information can be used to ensure privacy features. Therefore, this paper proposes a novel approach for iris authentication that ensures privacy features and high recognition performance. The binary representation of the existing iris template is transposed to another new binary representation by computing it column-wise. The Genuine Acceptance Rate (GAR) depicted for the proposed method is 97.11%, while the False Acceptance Rate (FAR) is 0.01%. Compared to the existing research works, the proposed method demonstrates an Equal Error Rate (EER) of 0.59%. The theoretical and experimental proof with explanation is stated in the paper to confirm the privacy aspects.