Image Watermarking Using Luminance Averaging and Adaptive Prediction from Nearby Pixels

In this paper, we propose a new method to enhance the performance of digital image watermarking based on pixels modification. The proposed method is implemented in both watermark embedding and retrieving processes. In the embedding process, an averaging technique is applied to the luminance components of the host image in order to obtain a better fine tuned watermark signal strength. In the retrieval process, a new prediction technique is used to adaptively remove a pair of nearby pixels having the highest difference from the predicting pixel in order to obtain a better watermark accuracy. The experimental results in terms of PSNR and NC values of the proposed method are given and compared to the previous watermarking methods. In addition, the improved robustness against attacks from the proposed method are presented and compared.