Adaptive partial image watermarking based on averaged variance of neighbor pixels

In this paper, we propose a method to improve the performance of the digital watermarking based on the modification of image pixels. In the embedding process, the watermark bits are partially embedded into the blue component of the image pixels. The first criteria, derived from the averaged variance of neighbor pixels of the watermarking pixel, are used to determine whether a watermark bit is embedded or not. The second criteria, derived in the same way as the first one, are used in the watermark extraction process to identity positions of the watermarked pixels. The experimental results show the improved performance and robustness in term of NC, compared to the previous methods.