PolyCodes: generating cancelable biometric features using polynomial transformation

Cancelable biometrics is a template protection technique proposed to bridge the growing gap between ‘ biometrics for security ’ and ‘ security for biometrics ’. Various approaches have been reported in literature, yet transforming a biometric template for privacy preservation is a challenging task. The manuscript proposes a random polynomial based template transformation technique to generate protected biometric templates named as ‘PolyCodes’. These protected versions are privacy and discriminability preserving, revocable, and provide significant dimensionality reduction. Several polynomial functions with dimensionality reduction factor ranging from 65% to 90% are designed and evaluated for security and performance. Further, applicability of the proposed ‘PolyCodes’ is experimentally verified for many biometric modalities such as visible and thermal face, palmprint, palmvein, and fingervein.