Inverse mapping of SCS-Watermarked Data

Scalar Costa Scheme (SCS) watermarking is a practical version of Costa´s capacity achieving blind watermarking scheme. In this paper, inversion of SCS, that is the removal of the embedded signal, is investigated. For the noiseless case, where no attack is performed on the watermarked data, SCS watermarking can be inverted perfectly. For the case of an AWGN attack, an MMSE estimate for the original host-data is presented, which gives an optimum estimate of the hostdata prior to watermark embedding.