Watermarking and Content Protection For Digital Images and Video.

With the advent of the world-wide-web, e-commerce and broadband connectivity to the home, electronic distribution of digital multimedia including music, photographs and movies, has been growing rapidly. Although there are many advantages, these technologies can create a serious problem for copyright holders. Consequently, protection for multimedia distribution against piracy is needed. A part of the solution is to use a watermark indicating content owner or distribution route, which is transparently and robustly embedded into the material. This thesis discusses several problems in watermarking design, including the imperceptibility and robustness to a variety of attacks. In particular, the StirMark random geometric attack is difficult to counter. Subsequently, several robust and transparent watermarking schemes for images and video and also an image confidentiality protection scheme, are presented. The original contributions can be summarized as follows. First, a robustness improvement of an existing scheme, namely watermarking by complementary modulation, is introduced. One existing method based on this to combat the StirMark random geometric attack, is shown to be invalid. Next, a new valid method to combat this random geometric attack is presented. Other methods of robust image watermarking, which consider the image texture, are proposed. By using these schemes, the watermark can be embedded in either low, high or both textures. Afterwards, an oblivious or blind detection method is suggested. This oblivious method can be used in conjunction with the previously described methods or can be used also with other existing methods. The above approaches are extended into video watermarking. A method to increase the imperceptibility by exploiting the human visual sensitivity is also designed. Subsequently, a method to detect video scene characteristic to suit a particular embedding scheme is presented. Finally, an efficient method to protect image confidentiality based on camouflaging and selective encryption, is introduced