Robust Video Watermarking Using Image Normalization, Motion Vector and Perceptual Information

摘要

This paper proposes a video watermarking algorithm robust against geometric distortions, several attacks of signal processing and intentional common attacks for video. Image normalization is used to get geometric invariant feature of video frames. Watermark embedding and detection process are carried out in the Discrete Cosine Transform (DCT) domain. Watermark energy is computed adoptively using perceptual information and motion vectors, to get major Watermark imperceptibility and robustness. Watermark Imperceptibility is evaluated by conventional PSNR and perceptual video quality measurement, taking sufficiently good visual quality. Computer simulation results show the watermark robustness to common signal distortions such as contamination by noise, JPEG compression, geometrical distortions and video common intentional attacks: frame dropping, frame swapping and frame averaging, among others.