Abstract
In our previous work, we proposed a biometric authentication method using a secure imaging system that was based on compressed sensing. In this approach, although we can acquire an encrypted vein image, the verification process requires the restoration of the raw finger-vein image. To address this issue, we propose an improved authentication method that we can verify alternate biometric features from which it is difficult to restore the original finger-vein image by introducing the permutation matrix for randomizing the object signal. Numerical simulations show that our method has favorable accuracy, although it exhibits a slightly degraded accuracy in comparison with that of the conventional method that uses a raw finger-vein image.
