An Automated and Robust Image Watermarking System Using Artificial Intelligence/Machine Learning Including Neural Networks
Gunjan Ahuja, Onkar Mehra, Muskan Aggarwal, Jashn Tyagi · 2024
In this study, we describe a CNN-based method for watermarking images. Embedding and watermark extraction are his two phases of the process. A CNN model is trained in the embedding phase to embed the watermark in the image. This ensures that the watermark is imperceptible to humans while also being resistant to image processing manipulations and attacks. The extraction phase uses another CNN model trained to accurately identify and remove watermarks from updated images. Experimental results show that the proposed method is successful, with good embedding capacity and robustness. CNN-based solutions provide automatic and efficient watermarking for large-scale applications to help protect digital images and enforce copyright.