Optimal parameter selection for image watermarking using MOGA

Abstract

The notion of the proposed methodology is to optimize multidimensional nonlinear problem of conflicting nature that exists among imperceptibility and robustness in image watermarking. The methodology exploits the potentiality of Multi-Objective Genetic Algorithm (MOGA) in searching multiple non-dominated solutions lying on the Pareto front. The characteristics curve of the image are then analyzed and the most appropriate solution is selected using a merit function defined over evaluation measures. The efficacy of the suggested method is demonstrated by reporting the resultant watermarked images and restored watermarks extracted from their mean and median filtered versions.