Modeling of Printing-and-digitalization Process - Lyon, France - Université Lumière Lyon 2

Université Lumière Lyon 2

Entreprise vérifiée

Lyon, France

il y a 2 semaines

Posté par:

Sophie Dupont

beBee Recruiter

Description

Modeling of Printing-and-Digitalization process using generative methods:

Réf

-
ABG-118557

Stage master 2 / Ingénieur
Durée 6 mois
Salaire net mensuel environ 600€650€ 30/11/2023
Université Lumière Lyon 2
Lieu de travail
Lyon Auvergne-RhôneAlpes France
Champs scientifiques
Informatique

Mots clés

Generative learning, image processing, image features, printable unclonable codes
Établissement recruteur:

Founded in 1973, Université Lumière Lyon 2 welcomes nearly 30,000 students on its two campuses, ranging from undergraduate to doctoral level.

As a university of literature, languages, and human and social sciences, it is comprised of 13 teaching units spread over four main areas of teaching and research.

With 33 laboratories and four research federations, which cover the areas of literature, languages, and human and social sciences (LLSHS - Lettres, Langues, Sciences Humaines et Sociales), Université Lumière Lyon 2 bases its approach on innovation, interdisciplinarity, partnership and an international outlook.

Through the projects developed and coordinated by its 1000 researchers, the university would like to enable communication and discussion between the human and social sciences, on one hand, and the hard sciences, on the other, as well as to put research at the centre of current societal and scientific challenges.

Université Lumière Lyon 2 has a strong focus on international cooperation and currently has agreements with 350 institutions throughout the world.

International students, whether part of an exchange programme or otherwise, account for more than 15% of the overall student body.

Description:

Context of the study

The nowadays challenges are the fast, reliable, and cheap detection of faked packaging and documents. Due to development and broad availability of highquality printing and scanning devices, the number of forged or counterfeited products and documents is dramatically increasing. Therefore, different security elements have been suggested to prevent this socioeconomic plague. One of the most promising and cheap solutions is the use of Copy Detection Patterns (CDP) [1]. A CDP is a maximum entropy image, generated using a secret key or password, that takes full advantage of information loss principle during printingandscanning process. Such an unpredictable pattern is highly sensitive to distortions occurring inevitably during production (printing), verification (scanning) and reproduction (duplication) processes.
The large variability of manufacturing chains (variability of printer technologies, paper substrates, etc.), and the variability of fake techniques make the authentication detector less reliable: as we need to find a tradeoff between the acceptance of a few fakes as authentic copies and the rejection of originals acquired with lower resolution.
The project

- _

TRUSTIT:

Theoretical and practical study of physical object security in real world use cases_ aims to explore the potential offered by deep learning methods in the context of CDP secure printing from the verifier's point of view.

During this internship we will work on learning a surrogate representation of the degradations added during printing process to the CDP.- The printable unclonable codes are printed at the limit of physical resolution of printing devices, therefore the considered authentication systems take a full advantage of Printing-Digitalization (PD) process.

This PD process is considered as physical unclonable function due to the optical and physical instabilities and is hard to be predicted.

In current stage, we need to print a big quantity of samples to learn the detectors. Nevertheless, the dataset construction process is very expensive and time-consuming process.

Additionally, the data collection process requires dedicated personnel and very strict procedures.- The generative neural networks (GANs) and latent diffusion models recently have showed their efficiency in data generation and style transfer [2,3].

Therefore, we want to use the advances on generative methods to propose the first method to model the PD process degradations.- This modeling is a very important step: it allows not only to predict the expected behavior of ink transfer on the support (paper) but also to ensure a better stability of authentication detectors by considering the influence of pixel neighborhoods.

Links and comparisons can be made with existing noise models such as the Kanungo approach [4].- Each printer and digitization device (in this step we will use a scanner) has its own signature.

In this internship, we will learn a surrogate representation of one pair printer-scanner using existing large dataset of printable unclonable codes [5].

We will experiment with different architectures of GANs and diffusion networks to identify the best approach for our task.- The pseudo-synthetic samples will be compared with real printed unclonable codes using some commonly used met