A copy attack resilient blind watermarking algorithm based on independent feature components

Abstract

Digital watermarking is an enabling technique to prove the ownership of copyrighted digital products. But its utility may be reduced greatly by copy attack which can cause ownership ambiguity. In this paper, a blind digital watermarking algorithm resistant to copy attack is proposed. By adopting the independent component analysis (ICA) technique, the most stable independent feature component (IFC) of the host image can be obtained in which to embed the watermark. Symmetrically, the watermark detection is just the investigation of the marked image's IFCs without the host image. In addition, the threat of copy attack can be mitigated, because no watermark can be detected from the copy attacked image's IFCs. Theoretical analysis and simulation results show that the algorithm proposed here can resist copy attack as well as the common image manipulations. © Springer-Verlag Berlin Heidelberg 2005.