| Digital watermarking is an enabling technology to prove ownership on copyrighted material, detect originators of illegally copies, and to monitor the usage of the copyrighted multimedia data. Currently, the most popular approach for digital image watermarking is the correlation-based spread spectrum technique. In this thesis, several novel techniques for image watermarking based on chaotic spread spectrum system are proposed. First, the chaotic spreading sequences are employed to spread out information signal instead of classical spreading sequences. Second, the chaotic parameter modulation is proposed to generate watermarks. The copyright information is embedded into the parameters of a chaotic dynamical system. The retrieval of copyright information is then a problem of parameter estimation from the extracted and possibly corrupted watermark signal. Different estimation methods are designed and the performances of them are analysed. It is shown that the chaotic parameter modulation based on mean value detection is not only superior to all the other estimation approaches, but it also has a better performance than the conventional correlation-based spread spectrum technique in terms of robustness, security and payload. |
