Research On Coverless Image Information Hiding Algorithm Based On Chaotic System

With the rapid development of network technology,network communication has become an indispensable part of people’s life.More and more information is transmitted on the public network,and the risk of eavesdropping is also increasing,especially some confidential information with high security level.Therefore,the security of secret information in network transmission has become an important issue in the field of information security.Coverless information hiding(CIH)is an important technology to achieve covert communication.CIH ensures the concealment of secret information transmission by mapping the relationship between image features and secret information or generating secret images based on generative adversarial network.However,the existing CIH methods still have the following shortcomings.For the mapping based CIH,they can hardly ensure both sufficient distinguishability and robustness to geometric attacks(such as rotation and flip).In addition,the completeness of secret information matching is rarely considered and the number of candidate images increases exponentially as the hiding capacity increases by one bit.For the generation based CIH,secret information is hidden in the generated image through the pretrained network structure,and thus there exists errors when restoring the secret information even without attacks.To address the problems that the mapping based CIH are difficult to ensure both the distinguishability and robustness,and the incompleteness of secret information matching at the same time,a novel CIH scheme based on chaotic key mapping is proposed in this thesis.In this scheme,multi-dimensional ring residual statistics are deployed to guarantee both the distinguishability and robustness against geometric attacks.In addition,a mapping mechanism is established by using LTCS between secret information and image features to ensure the completeness of secret information matching and reduce the required number of candidate images.Experimental results demonstrate that our scheme has superior performance in terms of distinguishability and robustness against geometric attacks,mapping completeness and the quantity requirement of mapping images.To address the problems that the generation based CIH has extraction error,and the mapping based CIH requires exponentially increasing number to sure mapping completeness when the length of secret information increased,a novel generation-mapping hybrid CIH scheme is proposed to improves the hiding capacity under the condition of lossless extraction of secret information.In this scheme,most of the secret information is hidden by image generation to reduce the quantity requirement of images for CIH.In addition,the residual information(extraction error)caused by image generation is then lossless compressed and a mapping mechanism is designed to further hide the compressed residual information for ensuring the lossless extraction of secret information.The experimental results demonstrate that our proposed hybrid algorithm not only improves the hiding capacity and greatly reduces the number of candidate images compared with the mapping based algorithm but also ensures that the secret information can be losslessly extracted compared with other exsiting generation based methods.In addition,the generated images have excellent visual effect which ensures the security of covert communication of secret information.