CNN-Prediction-Based-Reversible-Data-Hiding

Author:

Runwen Hu and Shijun Xiang

School of Information Science and Technology/School of Cyber Security, Jinan University Guangzhou, China

Description:

This version can only calculate the PSNR of images by using the proposed CNN-based predictor (CNNP) with expansion embedding and histogram shifting. The working environment is Windows 10, Python 3.7, PyTorch 1.6.0, and MATLAB 2019a. The work is based on the paper:

R. Hu and S. Xiang, "CNN Prediction Based Reversible Data Hiding," in IEEE Signal Processing Letters, vol. 28, pp. 464-468, 2021, doi: 10.1109/LSP.2021.3059202.

Folder description :

"standard_test_images": This folder contains four standard images used in this paper. Other images come from ImageNet.

"model": This folder contains the proposed CNN-based predictor.

"model_parameter": This folder contains the parameter of the proposed CNN-based predictor.

Usage:

python main.py [size] [model] [folder] [length]

size: The size of the images, which is set to 512*512 in this letter.

model: The saved model parameters.

folder: The place of the image to be predicted.

length: The length of the data to be hidden.

After using this program, the PSNR of the watermarked image by using the proposed CNNP-based reversible data hiding method will appear.

Example:

(1)For histogram shifting:

python main.py -size 512 512 -model .\model_parameter\model_state.pth -folder .\standard_test_images -mode histogram_shifting -length 10000

(2)For expansion embedding:

python main.py -size 512 512 -model .\model_parameter\model_state.pth -folder .\standard_test_images -mode expansion_embedding -length 10000