Dissertation
Dissertation > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > General issues > Security and confidentiality > Encryption and decryption

Video Watermarking Research Approaches Based on Human Visual Attention

Author ChenMengZe
Tutor LvYingHua
School Northeast Normal University
Course Computer Software and Theory
Keywords video watermark information hiding visual attention saliency map
CLC TP309.7
Type Master's thesis
Year 2011
Downloads 42
Quotes 0
Download Dissertation

Video watermark is a hot research area of the digital watermarking techniques. In the information society, the rapid increment of consumptive digital videos, i.e. VCD, DVD and other types of multimedia data, gives people enjoyment and convenience that they never had before. And, at the same time, the protection of the digital production’s property right and the security of information become an important problem, which makes the technology of video watermark much more necessary and important for the protection of property rights of the digital productions. Compared with the static images, videos include much more redundant information because the characteristic of time is inducted. So videos are more suitable for information hiding.In this paper, the basic conceptions of digital watermark are introduced at first, and then the characteristics of video watermark are showed. The main content of our research is focused on the video watermarking based on human visual attention model. The conceptions and foreground of visual attention are introduced in detail. At last, an algorithm of video watermark based on visual attention is proposed. Its design, implement and the analysis of the experiment results are discussed minutely. In the proposed algorithm, firstly, we get a static saliency map by using Itti’model , and secondly, a motion probability matrix is deduced from the frame difference method in order to build the dynamic saliency map. Finally, the global saliency map is obtained by the way of mean value integration, which is used to detect the regions of interesting (ROIs) in each video frame. There is a fact that human visual system attends some regions of the images or videos instead of whole of them. For example, people give more interest and attention to the moving objects, and in the medicine, the doctor is only interested in the lesion region. So hiding the watermark information in the ROIs can ensure the invisibility of the watermark to avoid people’s attention.Our experiment is based on the flat of MATLAB, a mount of experiments for the proposed methods of video watermark and steganography are performed. For the limitation of embedding time, only some consecutive frames are embedded. The results indicate that the proposed watermark algorithm not only is robust to familiar attacks, but also has better invisibility, robustness and security.

Related Dissertations
More Dissertations