In this paper, a new image watermarking framework for copiright protection is presented using image normalization, singular value decomposition (SVD) and dynamic stochastic resonance (DSR). The image is first normalized using the geometric moments and then an ownership message is embedded by employing the singular value decomposition. Finally, a reliable extraction process based on dynamic stochastic resonance is proposed. The DSR phenomenon helps in the extraction of the watermark information and improves the robustness of the system. A sequence of experiments and analysis are regulated to verify the effectiveness and the higher degree of robustness, especially for the geometric attacks. © 2018 IEEE.