A DWT-based Digital Video Watermarking Scheme with Error Cor(8)

Abstract. In this paper, a digital video watermarking algorithm is proposed. We present a novel DWT-based blind digital video watermarking scheme with scrambled watermark and error correcting code. Our scheme embeds different parts of a single watermark in

After extracting and refining the watermark, a similarity measurement of the extracted and the referenced watermarks is used for objective judgment of the extraction

fidelity and it is defined as:

(7)

which is the cross-correlation normalized by the reference watermark energy to give unity as the peak correlation [14]. We will use this measurement to evaluate our scheme in our experiment.

3. Experimental Results

To evaluate the performance of the new video watermarking scheme, several experiments have been done. They are: the experiment with various dropping ratio, the experiment with various number of frame colluded, the experiment with various quality factor of MPEG, and the experiment with various cropping ratio. Another DWT-based watermarking scheme which embeds an identical watermark in all frames is used to compare with the proposed scheme. A video clip with 1526 frames of size 352×288 is used in our experiment. The video consists of 10 scene changes. The NC values are retrieved when the watermarked video is under different attacks. The experimental results are described in details in the following.

3.1. Experiment with Frame Dropping

As a video contains large amount of redundancy between frames, it may suffer attacks by frame dropping. This experiment is aimed to examine the robustness of the scheme under attack by frame dropping. Different percentages of frames are dropped and obtained result is shown in Fig. 9.

Our scheme achieves better performance. It is because in each scene, all frames are embedded with the same watermark. This prevents attackers from removing the watermark by frame dropping. If they try to remove one part of the watermark, they need to remove the whole trunk of frames (i.e., the whole scene) and this would lead to a significant damage to the video. In addition, when frames are dropped, the error is only introduced to a corresponding small part of the watermark. For the DWT-based scheme (i.e., non-scene-based), however, the error is introduced to the whole watermark and it makes the performance worse.

The performance of the scheme is significantly improved by combining with a audio watermark, especially when the dropping rate of video frame is high. The improvement is increased with the dropping rate of the frame. This is because when the dropping rate increases, the error of the extracted watermark is increased and it significantly damages the watermark. The error correcting code from the audio watermark provides information to correct the error and overcome the part of the corruption of the video watermark, thus the NC values of the watermark is higher than