Different privacy protection mechanisms for biometrics have been suggested
in the literature since more than a decade. Many techniques are based on
template-level transformation without any secret key like fuzzy commitment
scheme and fuzzy vault or with the use of additional secrets, such as
cancelable biometrics.  In this presentation we will overview the security
properties which are usually targeted for such schemes.  We will emphasize
how an adversary may take benefit of the intrinsic properties of biometric
data to go beyond the usual theoretical security bounds. Nevertheless we
will also illustrate several schemes on different modalities for which we
achieved nice performances.

This talk is based on several joint works with various co-authors, in
particular Hervé Chabanne and Constance Morel from Morpho, and that have
been partially funded by European FP7 projects FIDELITY and BEAT.