A new semi-blind watermarking scheme based on DWT-subsampling

Aiming at the problem of poor robustness of existing digital watermarking algorithms to signal processing and geometric attacks, this paper presents a new semi-blind watermarking scheme based on discrete wavelet transform(DWT) and subsampling. For watermark embedding, the host image is decomposed into subimages by subsampling and after transforming these subimages via DWT, the embedding subgraph is obtained. Then its detail coefficients are transformed by singular value decomposition (SVD). Finally, the encrypted watermark is embedded into the singular values by watermark middle matrix. During the watermark extraction, the algorithm relies on the reference subgraph, achieving a semi-blind extraction. Experimental results demonstrate that the proposed scheme not only improve the Peak Signal-to-Noise Ratio(PSNR) value significantly, but also raise the robustness to JPEG compression, Gaussian noise and geometric attack etc.