A robust compressed sensing image encryption algorithm based on GAN and CNN

Abstract

In this paper, a robust compressed sensing image encryption algorithm based on generative adversarial network, convolutional neural network (CNN) denoising network and chaotic system is developed. Firstly, we use a sampling network to get the measurement of plain image. Second, the cipher image is obtained by scrambling the measurement through the Logistic-Tent chaotic system. After getting cipher image, the decryption party obtains decrypted measurement by inverse scrambling the cipher image, and then sends it to the reconstruction network to obtain decrypted reconstructed image. Finally, by using the CNN denoiser, the image quality and the visual expression of final decrypted image can be improved. In this scheme, the dual denoiser structure based on reconstruction network and CNN denoiser can effectively resist noise attacks. Besides, the proposed training strategy with noise injection can further improve the robustness of network. Experiments show our method has high reconstruction quality, efficiency, robustness and security.

KEYWORDS:

• Compressed sensing
• image encryption
• generative adversarial network (GAN)
• convolutional neural network (CNN)
• deep learning
• robustness

Acknowledgements

All the authors are deeply grateful to the editors for smooth and fast handling of the manuscript. The authors would also like to thank the anonymous referees for their valuable suggestions to improve the quality of this article.

Disclosure statement

The authors declare that they have no conflict of interest.

Additional information

Funding

This work is supported by the National Natural Science Foundation of China (grant number 41571417, 61802111, 61872125, 61871175), Science and Technology Foundation of Henan Province of China (grant number 182102210027, 182102410051).