Skip to Main content Skip to Navigation
Conference papers

Méthodes de réduction de la sensibilité à la base d'apprentissage en stéganalyse

Quentin Giboulot 1 Rémi Cogranne 1 Dirk Borghys 2 Patrick Bas 3 
LAGIS - Laboratoire d'Automatique, Génie Informatique et Signal
Abstract : Cover-Source Mismatch (CSM) has long been identified as one of the most important issues for hidden information detection. Recent works have shown that the CSM finds its roots mainly in the image acquisition and processing pipeline. This short paper builds on those recent research results by comparing two approaches for mitigating the CSM using the knowledge of the image processing pipeline. In particular, we have developed an efficient methodology for identifying the differeent processing pipelines used to generate a dataset. Based on this identification method we experimentally show that it is possible to set the steganalysis problem almost free from CSM using either a targeted approach, tailored to each and every processing pipeline, or diversifying widely the training dataset.
Complete list of metadata

Cited literature [18 references]  Display  Hide  Download
Contributor : Rémi Cogranne Connect in order to contact the contributor
Submitted on : Tuesday, June 11, 2019 - 4:46:47 PM
Last modification on : Sunday, June 26, 2022 - 4:36:33 AM


Files produced by the author(s)


  • HAL Id : hal-02152785, version 1


Quentin Giboulot, Rémi Cogranne, Dirk Borghys, Patrick Bas. Méthodes de réduction de la sensibilité à la base d'apprentissage en stéganalyse. Colloque GRETSI (Groupement de Recherche en Traitement du Signal et des Images), Aug 2019, Lille, France. ⟨hal-02152785⟩



Record views


Files downloads