A Novel Digital watermarking Approach under the DCT and SVM

The important direction of information hiding is digital watermarking technology. The traditional digital watermarking is advantage to against general attacks, but it is difficult to the geometrial distortions. This paper proposes a novel digital watermarking based on discrete cosine transform and support vector machine. Firstly, the block discrete cosine transformation is used for watermarking embedding, then the trained support vector machine is used to roughly correct the geometric parameters and the template matching method is used to correct the existing errors. Experiments show that the purposed algorithm is not only against traditional attacks such as noise, filtering and clipping, but also better robustness against geometrial distortions.