| The age of computer network has brought great advantages in the data acquisition, exchange and transmission, but the transparence of digital data also facilitates unauthorized copying and intentional tampering. As a potential and effective solution to these problems, digital watermarking, which embeds an imperceptible signal (the watermarking) into multimedia data such as image, text, video and audio without harming the host media's quality so as to protect them from illegal use, has attracted considerable attention in the past few years.This paper first briefly describes the state-of-art and the fundamental concepts of the digital multimedia watermarking techniques, then lucubrates on the video watermarking techniques in the compressed domain, and proposes two novel algorithms for video watermarking.The first oblivious algorithm is based on the texture direction of predicted frame errors. To ensure the video quality, the algorithm takes into account the mask effect of overlaying textures of the same direction and the characteristics of HVS to maximize the watermarking capacity without being seen, according to the sum of DC coefficients of reference block and predicted block, adaptively adjusts the embedding strength. By means of resynchronization scheme, the algorithm enhances further the robustness of the watermarking.The second scheme is a novel video watermarking algorithm based on the motion characteristics. In terms of the sign relationships of horizontal motion vectors and the ones of vertical motion vectors respectively, the algorithm first divides motion vectors into four different direction areas, then calculates the number of video objects in the frames using clustering theory and motion amplitude in different direction area. Mark the embedded frame using the characteristic, namely the number of video objects in the frame, and according to the human visual system (HVS) to embed the watermarking. Also, the algorithm further improves the robustness by resynchronization. |
PDF Full Text Request