Digital watermarking is being widely applied in various fields such as copyright protection of paintings, music, movies and documents together with broadcast monitoring, transaction tracking, copy protection and ownership authentication nowadays when digital multimedia is rapidly spread through the Internet.
Recently, the most actively studied digital watermarking covers image watermarking, document watermarking and video watermarking. In particular, image watermarking holds a high proportion.
Image watermarking can be divided into several kinds according to representation, embedding domain and objectives of applications ― visible watermarking and invisible watermarking from the aspect of representation; spatial domain watermarking and transformed domain watermarking according to embedding domain; robust watermarking and fragile watermarking for objectives of applications.
There are several indicators of the performance of watermarking including imperceptibility, robustness, capacity, security and computational complexity. In particular, imperceptibility and robustness are the two major features that are trade off, so a number of studies for an effective solution to it have been conducted.
Among them are approaches to raise imperceptibility by embedding watermarks into regions with the lowest human visual characteristics, approaches to get the highest robustness under given imperceptibility levels and approaches to solve single object optimization problems of empirical weighted imperceptibility and robustness by advanced evolutionary algorithms. However, all these have several drawbacks since they cannot correspond properly to arbitrary images and every attack.
Son Chol Min, a researcher at the Faculty of Applied Mathematics, selected blocks with lower human visual characteristics and embedded watermarks into their DC-transformed domains so as to raise imperceptibility. On the basis of it, he has presented a design method of an optimal blind watermarking algorithm which guarantees the highest imperceptibility and robustness for arbitrary images and against any types of attacks by solving multi-objective optimization problems between imperceptibility and robustness based on MOEA/D.
The simulation results showed that the imperceptibility of images watermarked by the proposed method was very high and this method outperforms previous ones in terms of robustness against various attacks.
You can find detailed information about this in his paper “Design of Optimal Blind Watermarking Technique based on MOEA/D” presented to the SCI Journal “IET Image Processing”.