Adaptive blind image watermarking using edge pixel concentration

In recent years many methods for image watermarking have been proposed to overcome the growing concern of copyright protection. The goal of all these methods is to satisfy the tradeoff between two important characteristics of robustness and imperceptibility. In this paper an adaptive blind watermarking method in the Contourlet transform domain is proposed. In this method we apply a two-level Contourlet transform on the original image. The first level approximate image is partitioned into blocks. Using a novel edge detection algorithm, important edges of each block of the approximate image are detected and the entropy of each block is also computed. Then by concatenating some portions of the second level subbands we form blocks. These formed blocks are transformed into DCT domain. Watermark is embedded by modification of the DCT coefficients. The severity of the embedding is controlled depending on the complexity of the corresponding block in the approximate image. For higher robustness against attacks, we embedded the watermark redundantly and used voting mechanism in extraction stage. Experimental results reveal that our method has high robustness and acceptable imperceptibility.