The pixel-wise adaptive Wiener filter watermarking

This paper proposes an efficiency prediction method for watermark extraction process in the amplitude modulation based watermarking method. The proposed prediction method using the neighbor´s pixel value is based on Wiener filter with neighborhood windows size of 3times3 pixels. The main advantage of our method which uses the estimation values based on the local mean and standard deviation of the watermarked image, is that a higher accuracy of the watermark recovery can be obtained, compared to the previous proposed methods. In addition, we assess the robustness of the embedded watermark against various common images processing such as Gaussian noise addition, JPEG compression standard and image cropping. PSNR and NC are used as indicators to show the effectiveness of the proposed method. From the experimental results, a better performance is achieved compared to the previous work proposed.