A modified robust embedding scheme for faithful watermark extraction

The watermarking of digital media for intellectual property protection, content authentication, and data hiding has been an intensive research area. An issue of considerable importance is how to embed an invisible watermark into a digital image so that it can be faithfully verified after innocent signal processing or unintentional attack. We propose a new watermarking algorithm that uses an “inversely-proportional” embedding rule for the low-magnitude transform coefficients. Experimental results show that this new technique is significantly more robust than the conventional transform-domain embedding rule