Video watermarking based on adjacent frames comparison in wavelet domain

In this paper, a new algorithm based on adjacent frames comparison in wavelet domain is proposed. In this algorithm, the embedding decision is made by comparing the LHx components of two adjacent frames. In the comparison process, a threshold is used to determine if the two components are similar enough. If they are, then the scrambled watermark image will be embedded to the LLx component of the frame that is named HostFrame. In the retrieving process, the LHx components of each two adjacent frames will be compared again to see if they are similar enough. If they are, then the watermark will be retrieved from the LLx component of the HostFrame. In the watermark retrieving process, the threshold should be the same with that in the embedding process to make sure the chosen frames contain watermark information. The results show that this algorithm has good frame-dropping and JEPG compression resistance.