Research Of The Algorithm In Temporal Synchronous Video Watermarking Based On SVD

As the development of the computer technology, Digital Videos are more and more widely used in people's life and the business applications. And the development of huge capacity access technology and the network facilitate the copy and share of digital videos greatly. On the other hand, the pirates can also wantonly modify the digital videos they got and share out illegally. This greatly infracts the copyright of the video owners. Many business applications (for example, the video advertisement) require the monitor of the digital video. Digital video watermarking can well protect the video's copyright and monitor the video using, with embedding the hidden watermark information to the digital video data. The digital video watermarking becomes more and more concerned.In the algorithm current researched of digital video watermarking, the algorithm based on three-dimensional transform is robust, but it has a problem of needing large amount of computation. Now the most concerned algorithm is the algorithm based on two-dimensional transform, but this should be considered about the temporal synchronization of the watermark between the inter-frames. This paper studies on the temporal synchronous digital video watermarking algorithm based on the two-dimensional transform.The matrix's singular value has good geometry invariance after the singular value decomposition. Firstly, this paper studies on a blind digital watermarking algorithm which is based on the singular value decomposition. In the watermark embedding, divide the video images into many blocks, and then decompose each piece of block with singular value decomposition, lastly embed the watermark into the singular value with fine-tuning the suitable singular value within the rage between the adjacent singular values. Because the orders of the singular values not change after adjustment, the detecting process does not require the original video data and the original orders of the singular values. So, this algorithm is a blind detecting algorithm, and the watermark is well invisible.The watermarking algorithm based on SVD can't resistant the temporal attacks because there is no correlated information between frames. To improve this shortage, this paper introduced a video watermarking framework for temporal synchronization and improves it, and then proposed a temporal synchronous video watermarking based on SVD (TS-SVD). The embedded watermark containing inter-frame related characteristics, and is resistance to most of the temporal synchronization attacks such as frame insertion attack. But, the original framework can't well resist against the attacks of frame drop and frame transposition. After analyzing the original framework's video feature exaction, this paper improves the framework by adding a resynchronization module. The improved framework has better ability of resisting against the frame drop attack.In this paper, experiments were taken to test the TS-SVD algorithm's robustness against geometric attacks and temporal synchronous attacks. Experiments show that the algorithm has a good invisibility and resistance to most geometry attacks, such as rotation, small rescaling, shifting, and MPEG compression. Experiments also show that the algorithm has a good robustness against temporal synchronization attacks, such as frame insertion attack, frame drop attack, frame transposition attack.