Research On Image Hiding Algorithm Based On Deep Neural Network

With the widespread and rapid development of computer network technology,the transmission,exchange,and sharing of information are mostly achieved through network communication.A large amount of secret information is transmitted through the network,which brings convenience to people's lives and work.Security risks.The information hiding algorithm can ensure that secret information is transmitted without being noticed by a third party,and improves the security of network communication,so it has been widely concerned by researchers at home and abroad.There are many carriers of information hiding such as text,audio,video,and image.Digital images are widely used in various industries due to their strong visibility and rich information expression.Therefore,this paper chooses images as the carrier of information hiding algorithms.That is,the image information is hidden.The secret image and the recovered secret image generated by the traditional image information hiding algorithm have a large gap with the original image,and cannot be adjusted according to the steganography analysis algorithm,so there are many limitations.In this paper,the research on image information hiding based on deep neural network is carried out for these problems.The main work and research results of this paper include:(1)An image hiding algorithm based on residual network is proposed.The algorithm refers to the image information hiding algorithm based on the predecessor based on the automatic encoder,and designs a new model structure based on the residual network.The carrier image is sent to the dense image through the shortcut(fast path)skipping the feature extraction layer The carrier image is used as a blueprint in the process of generating the dense image,which improves the transparency of the algorithm.From the analysis of simulation experiments,it can be seen that the algorithm can withstand the attacks of classic steganography analysis tools while generating high-quality encrypted images,and has strong security.(2)An image hiding algorithm based on residual and dense network is proposed.This work adds a dense connection to the preprocessing network on the basis of the hidden algorithm based on the residual network,so that the network can better extract the secret image features and reuse.Experiments show that this structure can effectively improve the quality of the secret image recovered by the algorithm and the dense image generated by the algorithm.(3)An image hiding algorithm based on generative adversarial neural network is proposed.In view of the security problems of the previous algorithm,the algorithm adds a steganographic analysis model to the previous model structure and designs the structure of the steganographic analysis model.The steganographic analysis model is trained against the previous model to improve Algorithm security.Experimental results show that the image hiding algorithm based on generative adversarial neural network proposed in this paper can effectively improve the security of the algorithm.