Abstract:
Biometric authentication has garnered considerable attention from researchers
in the past two decades. A biometric system is vulnerable to variety of attacks compromising the integrity of the process. One of the major attacks
is on the stored template to extract information about the individual. Unlike passwords and pins, The biometric data is permanent and once stolen,
cannot be canceled and re-generated. This has lead to the development of
so-called cancelable biometrics scheme which supports the cancellation and
re-generation of biometric templates. The schemes have already been successfully implemented for faces, palm prints, and fingerprints.
The idea of this work is to propose a new cancelable biometric approach
which has tentatively been named ”Expression Hash”. The idea is to hash the
expression templates with a set of pseudo-random keys which would provide a
unique code (expression hash). This code can then be serve as a template for
verification. Different expressions would result in different sets of expression
hash codes, which could be used in different applications. This will greatly
improve the privacy and security of applications. In case of compromise, the
existing code can be revoked and can be directly replaced by a new set of
expression hash code.
