Skip to Main content Skip to Navigation

Analyse et traitement des images dans le domaine chiffré

Pauline Puteaux 1
1 ICAR - Image & Interaction
LIRMM - Laboratoire d'Informatique de Robotique et de Microélectronique de Montpellier
Abstract : During the last decade, the security of multimedia data, such as images, videos and 3D data, has become a major issue. With the development of the Internet, more and more images are transmitted over networks and stored in the cloud. This visual data is usually personal or may have a market value. Thus, computer tools have been developed to ensure their security.The purpose of encryption is to guarantee the visual confidentiality of images by making their content random. Moreover, during the transmission or archiving of encrypted images, it is often necessary to analyse or process them without knowing their original content or the secret key used during the encryption phase. This PhD thesis proposes to address this issue. Indeed, many applications exist such as secret images sharing, data hiding in encrypted images, images indexing and retrieval in encrypted databases, recompression of crypto-compressed images, or correction of noisy encrypted images.In a first line of research, we present a new method of high-capacity data hiding in encrypted images. In most state-of-the-art approaches, the values of the least significant bits are replaced to achieve the embedding of a secret message. We take the opposing view of these approaches by proposing to predict the most significant bits. Thus, a significantly higher payload is obtained, while maintaining a high quality of the reconstructed image. Subsequently, we showed that it was possible to recursively process all bit planes of an image to achieve data hiding in the encrypted domain.In a second line of research, we explain how to exploit statistical measures (Shannon entropy and convolutional neural network) in small pixel blocks (i.e. with few samples) to discriminate a clear pixel block from an encrypted pixel block in an image. We then use this analysis in an application to correct noisy encrypted images.Finally, the third line of research developed in this thesis concerns the recompression of crypto-compressed images. In the clear domain, JPEG images can be recompressed before transmission over low-speed networks, but the operation is much more complex in the encrypted domain. We then proposed a method for recompressing crypto-compressed JPEG images directly in the encrypted domain and without knowing the secret key, using a bit shift of the reorganized coefficients.
Document type :
Complete list of metadata
Contributor : Pauline Puteaux Connect in order to contact the contributor
Submitted on : Thursday, January 21, 2021 - 3:12:08 PM
Last modification on : Tuesday, March 15, 2022 - 12:55:45 PM
Long-term archiving on: : Thursday, April 22, 2021 - 7:25:32 PM


Version validated by the jury (STAR)


  • HAL Id : tel-03117770, version 1


Pauline Puteaux. Analyse et traitement des images dans le domaine chiffré. Cryptographie et sécurité [cs.CR]. Université Montpellier, 2020. Français. ⟨NNT : 2020MONTS119⟩. ⟨tel-03117770⟩



Record views


Files downloads