A wavelet-based robust watermarking algorithm of high credibility

Abstract

In the digital watermarking technology based on wavelet transforms, it is of practical significance to improve the robustness of the JND (just noticeable difference) model and to control the probability of false alarm when extracting the watermarks from the watermarking systems. In the paper, a new watermark embedding algorithm has been studied and several experiments have been done as follows: a wavelet-based blind watermark detecting algorithm is proposed, which can determine the alarm threshold based on statistical characteristics, thereby quantitatively controlling the probability of false alarm; the JND model is improved, which enhances the robustness of JND model; and most of the embedded watermark can be effectively located even when the image is significantly modified. The algorithm proposed here is especially beneficial to blind extractions.