Sécurisation des fichiers multimédia par tatouage numérique

Abstract

In this thesis, we present two contributions in the field of color image watermarking. In the first contribution, we propose a new color image watermarking technique, based on Hessenberg Decomposition (HD) and Singular Value Decomposition (SVD) in DWT Domain, which embedding a color image watermark into a host color image, both of the same dimension, provides better imperceptibility, high embedding capacity and exhibits high robustness against attacks. An embedding factor is used to control the trade-off between watermark invisibility and robustness. This factor is manually selected for each color image used. However, this selection does not necessarily guarantee the optimal solution and requires a lot of testing time. Experimental results demonstrate that this technique provides high-quality imperceptibility and strong robustness against geometric, non-geometric, and combinatorial attacks.(geometric..and. non-geometric at the same time). The second contribution is an improvement of our technique based on DWT-HD-SVD. By exploiting the Artificial Bee Colony (ABC) algorithm. On the one hand, this algorithm allows finding the optimal insertion factors in order to obtain the best performances in terms of invisibility and robustness and on the other hand, it facilitates the use of the watermarking technique by automating the process of selecting the optimal insertion factor for each test image. A comparison with the most recent image watermarking techniques highlighted its advantage in terms of invisibility, resistance to attacks and embedding capacity.