A robust scalable spatial spread-spectrum video watermarking scheme based on a fast downsampling method

This paper presents a robust spatial spreadspectrum (SS) video watermarking scheme with adaptive detection for H.264/AVC and SVC. We introduce a fast image size change method in DCT domain to obtain a lowpass and downsized version of the video frame for embedding, which protects the watermarking signal from being damaged by downsampling during SVC encoding. And then we embed watermark bits by SS method with exploiting the Just Noticeable Distortion (JND) model to dynamically adjust the embedding strength. The watermark will be spread to the original video after the upsampling of the embedded downsized version. Furthermore, we deduce a detection model based on a likelihood ratio test to realize adaptive detection and design a different detection strategy for the base layer of SVC-encoded video. Simulation results prove the validity of the detection model and show that the proposed watermarking algorithm has strong robustness against not only H.264/AVC-encoded I-frames, P-frames and B-frames, but also SVC-encoded I-frames and P-frames while preserving the perceptual quality.