Robust Image Watermarking Using Coevolutionary Multi-objective Evolutionary Algorithm

In this paper, we proposed a coevolutionary multi-objective genetic watermarking scheme based on the wavelet packet transform. Wavelet packet transform can be viewed as a generalization of the discrete wavelet transform and a best wavelet basis in the sense of some cost metric can be found within a large library of permissible bases. Coevolutionary multi-objective genetic algorithm is used to select an appropriate basis from permissible bases of wavelet packet transform and determine the number of wavelet coefficients for watermark embedding. To balance performance of watermark robustness as well as image quality, watermark embedding process is treated as a multi-objective optimization problem. The proposed scheme will generate a set of non-dominated solutions, where each solution associates with a best wavelet basis and the number of wavelet coefficients used to embed a watermark. Experimental results demonstrate that proposed method can increase the capability to resist image processing methods while keeping quality of the watermarked image acceptable.