Image Encryption Communication Model Based On Multi-Attacker Adversarial Neural Network

Image encryption communication based on adversarial neural network as a task of image processing and encryption communication is faced with many challenges.From the perspective of cryptography,it is required that information can be fully recovered after encryption,and that ciphertext must be secure in dangerous environments.From the perspective of image,image space carries its pattern information,which requires the network to have the ability to extract and encrypt the spatial pattern information of image.Aiming at these problems,this paper proposes an image encryption communication model IECM based on multiple attacker adversarial neural network.The research work of this paper includes:In this paper,an image encryption communication model IECM is proposed based on generative adversity-network(GAN)and auto-encoder.The model includes sender,receiver and several attackers.Using a general loss function,this paper designs four IECM with different types of attackers,and simulates the attack of ciphertext by different types of attackers in different environments.For the original IECM model obtained above.In this paper,two aspects of performance optimization are explored,one is to add noise training to IECM,and the guidance model focuses on the image spatial pattern encryption learning to improve the robustness of the model.Another is to try to get IECM to learn challenging OTP encryption algorithms.For each IECM model proposed in this paper,experiments are carried out on MNIST handwritten digital image dataset and compared with several adjoint neural network encryption models.By comparing the experimental results before and after,it is found that the bad communication environment helps IECM to learn more secure encryption algorithm;After noise training,IECM can effectively learn the spatial pattern information of image encryption and decryption so as to improve the robustness of the model.Although limited by a variety of factors are not bright,with OTP encryption algorithm IECM to a certain extent to learn.Experimental results show that the proposed model can effectively solve the challenge of image encryption communication,and verify the effectiveness of the proposed method.