Proposal of Blind Extractable Additive Video Watermarking Method

To improve the quality of stego video, it is necessary to reduce the total amount of information embedded in the video as a watermark. This paper proposes two techniques. The first technique is distributed watermarking. A watermark is embedded in some of the blocks. The same watermark is embedded in multiple frames. By time averaging the extracted watermarks, error can be reduced. The second technique is detection signal embedding. In order to detect stego frames, a detection signal is introduced. This allows the watermark to be embedded in any frame. In addition, resistance to frame removal attacks can be provided. We embedded a watermark in videos and evaluated its performance using normalized correlations of the extracted watermarks. We also evaluated the quality of the stego video. The results showed that compared with the DCT-DWT method, the performance of the proposed method was better under different types of attacks, including compression attacks and frame removal attacks. In addition, the image quality was high with a PSNR of over 41 dB.