Research On Audio-Visual Feature-Based Digital Watermarking

With the development of computer network and multimedia techniques, the copyright protection and information security becomes a hot issue. Digital watermarking technology, which embeds additional information into original digital data, is one of the most effective methods to verify the ownership and authenticity of information. Digital watermarking has already been widely used for still images. However, many new watermarking schemes are proposed for other types of digital multimedia data, such as audio, video, text and 3D meshes. Due to the fact that original information is the carrier of digital watermarks, if there's a probability of exploiting the audio and visual features of multimedia data, a watermarking technology based on these features would boast of better performance and practical value. Authentication and copyright protection for multimedia act as the most important application drives in the development of digital watermarking technology. Therefore, there is a focus on applications for authentication and copyright in this paper.Basics about video watermarking are covered at the very beginning of this paper, then the key issues for digital watermarking are proposed. After analysis on some important parameters and probable attacks on watermarking scheme, some typical digital watermarking algorithms for single multimedia like audio or video are discussed.Next, MPEG-2 video codec is introduced. Based on both the static and dynamic features of video as well as MPEG-2 scheme taken into account, experiments are carried out in these two aspects followed by detailed analysis of the performance. In regard to temporal synchronization, a simple but feasible way is put forward. In chapter 4, the typical phase coding algorithm is studied. Experiments show that this algorithm doesn't perform well under the requirement of robustness. After in-depth study of audio features, an effective algorithm for audio watermarking based on audio power is proposed.Today, the available watermarking technology concentrates on single media like audio or video. But a typical multimedia stream consists of both video and audio data. Watermarks for single media might not be able to guarantee the integrity of the multimedia data. Therefore, it is necessary to watermark the combined audio and video to protect against alteration of one of the components without interfering the other. In chapter 5, two watermarking schemes for different applications are proposed. The experiments indicate that the first scheme meets the robustness requirements when frame dropping, swapping or inserting is imposed. The second watermarking scheme is semi-fragile and can be applied to content authentication of audio-visual data in journalism, commerce and law.