Research On Digital Watermarking Technology Based On H.264/SVC

The improvements of network infrastructure and advances in video coding technology breeds a series of video applications including video telephony, VOD, high definition TV broadcasting, et al. In 2007, in order to make compressed video stream better adapt to various network environments and user terminals, a video coding standard SVC (Scalable Video Coding) was developed by the JVT (Joint Video Coding). SVC enables compressed video to decode with different temporal, spatial and quality resolution, which provides a simple, flexible and effective solution to network video transmission and has great potential of applications. However, with the increasingly rampant illegal usage of video, how to efficiently realize safe and reliable distribution of SVC encoded video and how to efficiently protect the vital interests of the copyright owner present new challenges to researchers. Digital watermarking is expected as a powerful technical tool to solve the above problems by embedding an imperceptible, yet detectable signal into the video sequence. This paper made in-depth research of SVC-based watermarking technologies and proposed two watermarking schemes, which is a beneficial exploration for providing protection to SVC encoded video. The main work of this paper is as follow:Firstly, we proposed a robust SS (Spread Spectrum) watermarking scheme for SVC by employing a fast image size change method in DCT domain. The watermark bits were embedded in the downsampled version of the original video, and the embedding strength were dynamically controlled by two JND values. One JND value was propagated through downsampling and the other was calculated in the downsampled version directly. We chose the smaller JND to ensure that both of the original video and the downsampled video would not produce noticeable distortion to human eyes. The spatial base layer and the enhancement layer used different method for watermark detection. Experiment results show that our algorithm has small impact on the quality of decoded video, and a high accuracy rate of watermark detection is gained for both of spatial enhancement layer and base layer.Secondly, this paper realized a high capacity scalable watermarking algorithm inside the reconstruction loop of SVC encoder. Considered to decrease the bit rate, the secret bits were inserted by altering the last quantized AC coefficient in Zigzag scan order. Because the watermarking process was inside the reconstruction loop, the coefficients after embedding would take part in mode decision and prediction. The order of watermark bits insertion was the same as that of video frame encoding. Experimental results demonstrate our schemes outperform other video watermarking schemes in higher hiding capacity and less bit rate increase, especially under a large quantization step.