Optimum decoding and detection of multiplicative watermarks

This work addresses the problem of optimum decoding and detection of a multibit, multiplicative watermark hosted by Weibull-distributed features: a situation which is classically encountered for image watermarking in the magnitude-of-DFT domain. As such, this work can be seen as an extension of the system described in a previous paper, where the same problem is addressed for the case of 1-bit watermarking. The theoretical analysis is validated through Monte Carlo simulations. Although the structure of the optimum decoder/detector is derived in the absence of attacks, some experimental results are also presented, giving a measure of the overall robustness of the watermark when attacks are present.