Robust image watermarking algorithm based on residual neural networks
Yunfei Wang, Gehao Lu · 2025
With the development of the Internet, information carriers and communication methods have become diversified. As an important carrier for obtaining information, images are easily copied and tampered with during the communication process. Digital watermarking technology is an effective technology to protect image copyrights. In recent years, with the development of artificial intelligence, scholars have begun to apply deep learning technology to digital watermarking technology. This article proposes a digital watermarking technology based on residual neural networks, which consists of host image processing network, watermark processing network, watermark embedding network, and watermark extraction network. Experiments show that the watermarking algorithm can effectively resist common attacks on images, has good robustness, and the image will not change greatly after embedding the watermark, so it has good invisibility.