Secure digital image watermarking based on SVD-DCT

Abstract

Singular Value Decomposition (SVD) has been used as one of the best transform for implementing robust digital watermarking. In some SVD based watermarking techniques, eigenvalues of the watermark are embedded into eigenvalues of the cover image and the eigenvectors are used as key parameter. This category of methods is not capable to provide the secure watermarking technique as the original eigenvectors have been used in extraction process. Apart from that, the other SVD based watermarking which are considered as secure techniques are not enough imperceptible. In this paper, two new secure methods of non-blind image watermarking is performed. These methods are implemented using modification on singular values of the host data by embedding DCT coefficients of the watermark image. Due to the nice properties of DCT as a reversible transform and taking the advantages of same variation as the Singular values change, these coefficients of watermark has been used as the information which is embedded to provide the secure schema. Beside enough robustness against large amount of attacks, the best quality for watermarked image can be achieved by altering just brightness instead of changing the local variation of the signal. © 2008 Springer-Verlag.