Effective hill climbing algorithm for optimality of robust watermarking in digital images

Due to the explosion of data sharing on the internet and the massive use of digital media, especially digital images, there is great interest by image owners in copyright protection. The genetic watermarking methods were previously shown to optimize the conflicting requirements of robustness and invisibility. However, the genetic watermarking methods have limitation in considering perceptually significant or non-significant regions in the selection process, so they do not always offer better imperceptibility. In addition, the computational resource required by Genetic Algorithm (GA) is high when comparing it to other heuristic methods. Thus, the current study is focused on an optimization-based Dither Modulation watermarking scheme for digital images in a more efficient and effective manner. The watermark imperceptibility and robustness are taken into consideration at the same time. A hill climbing algorithm, which has a simple computational process, is employed for optimizing these two conflicting requirements. Since, Peak Signal-to-Noise Ratio (PSNR) may not be an effective imperceptibility measure presented previous genetic watermarking methods, Watson's perceptual model is employed to quantify the watermarked image distortion as it is consistent with Human Visual System (HVS). Several commonly used watermarking attacks are considered in the optimization process. Experimental results demonstrated that the proposed algorithm is robust and more time efficient than the previous GA based methods.