A Reversible Data Hiding Method Based on Super-resolution Reconstruction Network
Chunquan Yang, Xichuan Hu · 2023
With the emergence of deep learning, new possibilities have arisen for reversible data hiding. In this paper, we propose a reversible data hiding method that leverages an image super-resolution reconstruction network. The method employs a specific neural network as the "key" for both data embedding and extraction. Firstly, the original image is input into the neural network model to obtain a higher-resolution reconstructed image. Then, a portion of the pixels is used to store the original image while the rest of the pixels are used to embed the secret data, resulting in a carrier image. Finally, the original image can be extracted from the carrier image by feeding it back into the neural network model for reconstruction. The hidden data can then be extracted by comparing the difference between the reconstructed image and the cover image. The proposed method takes into consideration both the embedding capacity and image quality, providing a good embedding capacity while ensuring a certain level of image quality.