| With the development of multimedia technologies and the popularization of Internet, multimedia content becomes an important part of peoples'work and entertainment. Though digitalization brings ease to our daily life, great challenge has taken to the copyright protection of multimedia works due to its properties of easing copy and redistribution. More and more multimedia creators are not intent to open their works because of lacking efficient copyright identification technologies and traitor tracing mechanism, which would impede the development of digitalization and the application of multimedia.As a branch of digital watermarking technology, digital fingerprinting is used to trace the leakage of the multimedia content. It becomes a hot topic in recent years. Most of current researches regarding to digital fingerprinting are based on theoretical application, users accommodated in fingerprinting system are usually small. However, with the increasing of user scale and the requirements of video fingerprint application, the fingerprinting system calls for improvements, such as robustness, collusion-resistance, and real-time processing, which are problems in current research. Traditional fingerprinting embedding strategies consider improving robustness or alternatively real-time processing. Large-scale fingerprinting application requires the algorithm meeting both of them. Therefore, studying a new fingerprinting embedding strategy which is able to satisfy large-scale applications is very necessary.To lower the computational complexity of fingerprint embedding, we propose a spatial-domain based robust real-time embedding algorithm in this paper. At first, the fingerprints are converted into an embedding template. And then, the template is superimposed to the host signal directly in spatial domain. The proposed embedding algorithm is simple and efficient, and it could be used in various types of multimedia data. Though the fingerprint template is added in spatial domain, the efficiency is equivalent to that of the methods in DFT (Discrete Fourier Transform) domain. Therefore, the robustness can be maintained. Comparing with the traditional algorithms in transform-domain, the proposed algorithm performs much better in real-time processing. This is because that no transformation is needed for video frames or images during the embedding process. The experimental results show that the proposed algorithm has many advantages in real-time processing and robustness. |
