A new binary watermark embedding method is proposed in this paper based on chaotic mixing and singular value decomposition (SVD) for copyright protection and authenticity. We used the chaotic mixing to mix binary watermark image and SVD transformation to extract the left orthogonal matrix of host image at the same time. Other than traditional watermarking algorithms based on SVD, the binary watermark bits are embedded in the left orthogonal matrix instead of the singular values (SVs) of the host image's sub-blocks, because the orthogonal matrices contain the texture and detail information of the host image, while the SVs only contain the energy information of host image. This greatly enhances the safety of watermark because the SVs can be easily replaced without much damage to the host image, whereas the orthogonal matrix cannot. Compared with another watermarking method based on modifying the orthogonal matrix, our method maintains the orthogonality of the modified orthogonal matrix. Experimental results showed that our method had higher robustness and was better in resisting normal attacks. The error ratios were reduced by an order of magnitude using the proposed method.

No abstract