Protecting Biometric Features by Periodic Function-Based Transformation and Fuzzy Vault - Transactions on Large-Scale Data-and Knowledge-Centered Systems - page 58

without the user's consent. Another problem is that unlike passwords, cryptographic
keys, or PINs, biometrics cannot be changed once compromised. In addition, a user
can be tracked by means of cross-matching when he/she uses the same biometrics
across all applications and the service-providers collude with each other.
Therefore, the security of biometric template has been emerging increasingly and a
lot of research has been done in this field. According to the authors in [3], there are
four properties that an ideal biometric template protection scheme should possess:

1. Diversity: the secure template must not be the same in two different applications;
therefore, the user's privacy is ensured.
2. Revocability: it should be straightforward to revoke a compromised template and
reissue a new one based on the same biometric data.
3. Security: An original biometric template must be computationally hard to recover
from the secure template. This property guarantees that an adversary does not have
the ability to create a physical spoof of the biometric trait from a stolen template.
4. Performance: the biometric template protection scheme should not degrade the
recognition performance of the biometric system.

In biometric template protection, fuzzy vault is considered as a popular method. It
binds a key with the biometric template and obtains the helper data for authentication.
The template is hidden in the helper data. However, there are some problems that
fuzzy vault encounters with. One of these stems from the reason that fuzzy vault can-
not provide the diversity and revocability properties. In this paper, that shortcoming is
made good by applying a feature transformation in a fuzzy vault scheme. Strictly
speaking, the main idea for the marriage of fuzzy vault with feature transformation
was introduced in a few recent proposals. Nevertheless, majority of which focus on
two biometrics modalities, fingerprints [4, 5, 6], and iris [7, 8]. The transformations
for face based fuzzy vault scheme are rare and very complicated. Therefore, this paper
will present a hybrid scheme of face based fuzzy vault and feature transformation.
Our proposed transformation is not only simpler but also suitable for many kinds of
biometrics modalities. The face biometric templates are protected by hidden in the set
of chaff points generated by fuzzy vault scheme. Besides, these templates are able to
be changed or revoked if the owners have suspicious about being tracked or stolen.
Our experimental result will show that the newly proposed scheme guarantees the
revocability property with an acceptable error rate.
The structure of this paper is organized as follows. Section 2 provides a brief re-
view of related works. The details of our proposed scheme are described in Section 3.
Following them, the evaluation is discussed in Section 4. At last, Section 5 provides
the conclusion and future works.

2 Related Works

Biometric template protection is an important issue in a biometric system. Biometric
template of a person cannot be replaced or used again once it is compromised. In [9],