| Digital watermarking is considered a possible solution to the problem of protecting digital material. Many techniques have been proposed for watermarking images and videos. One major difference between these techniques is the actual embedding schema employed to embed data inside the host image; however, the literature lacks an effective comparison of the different embedding approaches.; The work presented in this thesis introduces and highlights the technical challenges associated with watermarking. Our concern is to determine the best embedding schema for a watermarking algorithm and to use the Human Visual System (HVS) features for automatic optimization of watermarking embedding schema parameters. We first give an introduction to the watermarking systems, processes, properties, and applications and then an overview of the compression standards as well as a comprehensive survey of current watermarking techniques and approaches.; Two different embedding schemes are considered in this thesis. In the first method, a watermark is embedded into the host image and then the watermarked image goes through the compression process. In the second scheme, the cover image goes through the entire compression process and then a watermark is added to the compressed image. For both approaches the JPEG and the wavelet compression standards are employed. The issue of false negative rate and the fidelity level are addressed and we propose a mathematical method of using the human visual system (HVS) features as a solution to a related optimization problem. Also we suggest ways in which the performance of the watermarking algorithms can be improved.; Finally, through statistical analysis, we recommended a set of criteria for evaluating our proposed embedding schemes, as well as a mathematical method for using the HVS, which makes novel use of tool-sets not previously proposed. We close with a summary and conclusions and suggest future research directions. |
