Based on SVR Adaptive Watermarking Technique in DCT Domain

A new image watermarking based on the support vector regression (SVR) and human visual system (HVS) is proposed in this paper. According to the discrete cosine transform (DCT) suggested a new watermarking algorithm, human visual system is used to find the suitable strength where the watermark should be embedded and support vector regression is used to simulate human visual system to achieve maximum strength that is embedded into the original image. Besides, the proposed scheme can extract the digital watermark without the help of the original digital image. Different experiments are conducted to show the performance of the scheme under different types of attacks. The results show that the proposed scheme resulted in very good imperceptibility and high robustness. Especially the robustness against the JPEG compression is achieved for a high level.