Tatouage d’images basé sur des transformées discrètes entières

Abstract

This work considers image watermarking by exploiting the visual masks recently developed in the literature. The thesis firstly presents the state of the art on image watermarking techniques and their applications. This study has allowed us to develop two new robust blind image watermarking algorithms. The first algorithm is essentially dedicated for using the spread spectrum technique in the DCT and integer DCT domains along with an appropriate exploitation of the local features of the image to design the visual masks. However, the second algorithm is dedicated for exploiting the QIM watermarking technique in the 3D-DCT domain. The simulation results to evaluate the robustness and the objective measures carried out on the watermarked images attacked with different attacks, presented in this work show clearly the efficiency and robustness of the algorithms proposed in this thesis. Moreover, these proposed algorithms present an advantage of reduced complexity compared to those of the existing image watermarking techniques.