Towards a New Metric for Watermarked Image Assessment

Abstract

To evaluate the watermarking algorithm performance, one of the basic factors used for this purpose is the watermark imperceptibility, which reflects the degradation of watermarked image in compare to the original image. In this paper, the performance of the most commonly used image evaluation metrics, peak signal to noise ratio (PSNR), and structure similarity index (SSIM) are tested in high textured and low textured images, in addition a new metric which considers the texture mapping in assessing the watermarked images has been proposed. Experimental results show better estimation for the watermarked image quality when the texture masking is involved in image assessment in compare to traditional metrics.