A New Approach for Cancelable Iris Recognition

ABSTRACT The iris is a stable and reliable biometric for positive human identification. However, the traditional iris recognition scheme raises several privacy concerns. One’s iris pattern is permanently bound with him and cannot be changed. Hence, once it is stolen, this biometric is lost forever as well as all the applications where this biometric is used. Thus, new methods are desirable to secure the original pattern and ensure its revocability and alternatives when compromised. In this paper, we propose a novel scheme which incorporates iris features, non-invertible transformation and data encryption to achieve “cancelability” and at the sa me time increases iris recognition accuracy. Keywords cancelable biometrics, iris recognition, security, privacy 1. INTRODUCTION Biometrics is an emerging field of technology using unique and measurable physical, biological, or behavioral characteristics that can be pr ocessed to establish identification, to perform identity verification, or to recognize a person through automation. [1]-[5] Biomet ric characters can be anatomical—such as iris[4], face[6], finger image[7], hand, and voice, etc, or behavior , including signature, gait and typing rhythm, etc . Among these biometrics, the iris is tested to be one of the most reliable traits for positive identification. A typical biometric system includes three major steps: signal (image) acquisition, signal (image) processing and representation and pattern matching. A typical biomet ric identification procedure includes the enrollment stage and the authentication st age. During enrollment, the raw biometric data are inputted and processed in the feature extractor and the extracted template are th en stored into database. In the authentication stage, the same feature extraction procedure are implemented on the query biometric signal, and using pattern recognition methods, we check whether it is rela ted to its claimed identity in the database(Fig.1) However, biometric patterns are usually vu lnerable [8], which means once your biometrics are compromised, they could be compromised forever or be very hard to recover. When a biometric is compromised, the user has very