A New Robust Watermarking Scheme Based on Features Classification Tree

Digital watermarking technique is a common solution to image authentication and authorization. Numerous researches in this area use modification method to alter the pixel value or the coefficients of transform domain to succeed the watermarking purpose. In this paper, we propose a novel watermarking scheme using the concept of association rules to build a features classification tree and Chinese remainder theorem to embed the index of the tree instead of watermark's native values, and via the inverse process the watermark extraction is accomplished. Additionally, it has application ability to any watermarking techniques in the same principle of modification method. Our experimental results show not only the robustness against different image processing attacks, but also the enlargement of capacity to progressively accept the gray scale watermark.