Robust Video Watermarking using Multi-Band Wavelet Transform

This paper addresses copyright protection as a major security demand in digital marketplaces. Two watermarking techniques are proposed and compared for compressed and uncompressed video with the intention to show the advantages and the possible weaknesses in the schemes working in the frequency domain and in the spatial domain. In this paper a robust video watermarking method is presented. This method embeds data to the specific bands in the wavelet domain using motion estimation approach. The algorithm uses the HL and LH bands to add the watermark where the motion in these bands does not affect the quality of extracted watermark if the video is subjected to different types of malicious attacks. Watermark is embedded in an additive way using random Gaussian distribution in video sequences. The method is tested on different types of video (compressed DVD quality movie and uncompressed digital camera movie). The proposed watermarking method in frequency domain has strong robustness against some attacks such as frame dropping, frame filtering and lossy compression. The experimental results indicate that the similarity measure before and after certain attacks is very close to each other in frequency domain in comparison to the spatial domain.