Robust watermarking based on optimal synchronization signal

Watermark synchronization is the key to the success of watermark extraction. In this paper, a new robust watermarking method based on optimal synchronization signal is proposed. Firstly, a new image block order and watermark embedding order are introduced. Based on this, the optimal synchronization signal is analyzed by using Markov chain, and the conditions and probabilities of the synchronization signal detection are given. Then, the watermark is embedded into the third singular value and the subsequent singular values in DWT-SVD domain, which can balance the visual quality and robustness. In watermark extraction, the geometric attack parameters are first estimated by using the synchronization signal and a similarity function. Then, the watermarked image is restored and performed to extract the watermark by relative relation between singular values. The experimental results show that the proposed watermarking method achieves excellent robustness to both image processing attacks and geometric distortion attacks.