Non-blind Arnold Scrambled Hybrid Digital Image Watermarking Scheme based on Differential Evolution and DnCNN

Abstract Watermarking is to place a message invisibly in the media without damaging it. Transform (frequency) domain based watermarking has received more attention due to its advantages over spatial domain. Many algorithms which utilize SVD, have a false positive problem. Therefore, this study propose a scheme that along with the robustness of the cover image to the various attacks, maintains the imperceptibility of the watermarked image. To achieve these goals, we use three transforms along with Differential Evolution algorithm. Also, to obtain the best multiple scaling factors, we modify the unique values of SVS of the low resolution approximate image which is obtained in the embedding process. To further increase the security of the proposed method, we scrambled the cover image using Arnold’s transform. We also did not use the U and V diagonal components in the embedding and extracting steps to prevent false positives.