Abstract—Based on the observation that low-frequency DCT

coefficients of an image are less affected by geometric processing,

we propose a new blind MPEG-2 video watermarking algorithm

robust to camcorder recording. Both the embedding and

extraction of the watermarks are done in the compressed domain.

The mean of the low-frequency DCT coefficients of the video is

temporally modulated according to the information bits. To

reduce complexity, we embed watermark in the prediction error

signals of the B-frames. Experimental results show that the

proposed scheme achieves high video quality and robustness to

camcorder recording, trans-coding, frame-rate conversion, and

geometric attacks on common test videos.