Multiple video Zero-Watermarking based on 3D DFT to resist geometric attacks
Huaiqiang Zhang, Jingbing Li, Chunhua Dong · 2012
In this paper, we propose a multiple watermarks technique for video sequences. The proposed algorithm is based on the three-dimensional discrete Fourier transform (3D DFT). The video sequences are divided into different shots and key frames are selected from those shots. The 3D DFT is performed on these key frames and the coefficients of the 3D DFT are got. The sign sequence of 3D DFT coefficients is used as feature vector. Then three binary sequences are used as the watermarks and combining the concept of third-party, we describe how to embed and extract the watermarks. Experimental results show that the algorithm bears desirable robustness on geometric attacks.