Vincent Despiegel

Morpho, France

Vincent Despiegel received the Agrégation de Mathématiques degree in 2004, is a former student of the École Normale Supérieure de Lyon, Lyon, France, and received the Ph.D. degree in 2007 from the Université de Grenoble, Grenoble, France, on the study of Hyperelliptic curves and their use in the generation of substitution boxes for cryptographic applications.

Since 2007, he has been a research and development staff member at Morpho, France, within the Biometric Research Team. From 2007 to 2011, he worked mainly on fingerprint algorithms improvement. In particular, he was involved in the European FP7 integrated project TURBINE (TrUsted Revocable Biometric IdeNtitiEs, 2008-2011) and worked on template protection and fingerprint templates binarization. Since 2011, he is the leader of the research team dedicated to face detection and tracking. His research interests include cryptography, image processing and pattern recognition dedicated to biometry.

Talk 1 : Large scale system in biometrics

Identifying and verifying someone's identity in a fast, and accurate manner is a growing need in our modern society. It is growing both in scale and in depth. In scale, because most of the latest identity schemes rolled out by countries involve either the general population, or at least a very large one, e.g. the Indian UID project. In depth, because rights granted based upon automatic biometric identification have an impact on how we live our lives : the right to vote, the right to welfare, to right to travel overseas. The roll out of these massive identification / identity management systems is occurring right now, and although many issues have already, to some extent, been addressed, such as scalability, performance, robustness, ergonomic or inclusiveness, there are many pain points still out there. In a short presentation, the speaker will describe some of the existing solutions, challenges, areas of research of the domain through the lens of the three prominent modalities today : fingerprint, face and iris.

Talk 2 : Template protection and biometry

Today, when comparing classical fingerprint matching and more constrained algorithms, like after binary quantization for biometric privacy protection purpose, there is an important gap in terms of performances. Performances of the latter solutions still need to be improved to decrease this gap. The main difficulty comes from the fact that fingerprint

captures of the same trait give very different minutiae sets with possibly small overlaps and very different distortions

among the different part of the images; and comparison of a stored reference with the fresh captured fingerprint data has to take into account local variability. After giving an overview of classical template protection techniques, the presentation will focus on binarization issues for different biometric traits.