Abstract
Digital watermarking plays an important role for copyright protection, content authentication, and annotation of digital multimedia including audio, video, and still image. Because digital video content providers are concerned with copyright issues, a proper watermarking scheme for digital video is demanded. The watermark can be either embedded in the uncompressed video sequence or the compressed video sequence. Since many available video sequences are compressed, our proposed algorithm embeds the watermark in the frequency domain for the compressed video sequence. If we want to embed the watermark in the frequency domain, the additional decoding operations to DCT (Discrete Cosine Transform) coefficients for each picture are still too expensive. The low-complexity embedding scheme should be considered. On the other hand, imperceptible watermark based on HVS (Human Visual System) is also an important consideration. In this thesis, we proposed an effective watermarking scheme for digital video. In our work, we apply a MPEG-1 video as an experiment. Our proposed watermarking scheme is to embed a meaningful logo to the high capacity blocks of Watson’s DCT-based visual model of the Y plane of the I picture of a video sequence. The proposed algorithm directly substitutes the VLC (Variable Length Code) without inverse quantization. It can reduce computational time-complexity. The proposed watermarking algorithm is also capable of detecting satisfactory watermark for the attacks of video compression.