Research And Design Of Information Hiding Mechanism Of Embedded Platform Based On Deep Learning

In today's interconnected under the time background of all things,people place more than ever the security of the information transmission,and the embedded platform has a growing demand for covert communication of information.Information hiding technology can hide secret information in the carrier without being perceived or detected by the third party,which not only protects the information content but also protects the security of the information transmission process.At present,the most widely spread digital media has opened up a new development space for the information hiding method based on the image.Commonly used information hiding method based on the image is mostly by modifying the image data and to hide the secret information,so it is difficult to resist the detection of steganographic analysis using deep learning.In recent years,due to the deep learning algorithms such as generative adversarial networks and convolution neural network have made remarkable achievements in the field of the image,so it has become the focus of research to improve information hiding technology by using deep learning algorithms and ideas.This paper studies and designs an embedded platform image information hiding mechanism based on deep learning.The mechanism satisfies the need of hidden communication,and improve the information hiding mechanism analysis of resistance and the quality of the secret image.The main research results of this paper include the following aspects:(1)Aiming at the problem that traditional image information hiding algorithms are difficult to resist steganography analysis due to the embedding of information by modifying carrier information,a generative image information hiding mechanism based on deep learning is proposed.Through the training of generative adversarial networks,the pseudo-natural image similar to the data set was generated,and the generated image was not modified.Covert information is transformed into a noise vector through mapping rules,which is used as the input to generative adversarial networks,and images containing covert information are directly generated.Then a convolutional neural network is trained using the generative model of generative adversarial networks for the extraction of hidden information.This mechanism has a strong capability in anti-steganography analysis and image forensics and has some improvement in the security of information hiding mechanism.(2)To deal with the issue of the low quality of the cryptic image in the image information hiding algorithm based on generative adversarial networks,this paper proposes an information hiding algorithm model based on Realness-DCGAN.The objective function is optimized by using the concept of distribution,and the fidelity distribution is introduced to guide the training of model generation more effectively.The improved information hiding algorithm model can generate higher quality cryptic images and improve the information recovery accuracy of the extractor.(3)Given the demand for covert communication in the information transmission process of the embedded platform,the information hiding mechanism based on deep learning proposed in this paper is applied to the embedded platform to realize the secure transmission of information.Through experimental tests show that the proposed mechanism of information hiding in the embedded platform with less running time and memory usage,satisfy the embedded platform resource-constrained conditions,can be used in the embedded platform.Based on the information hiding mechanism of experiments and tests,the experimental results show that the presented based on information hiding mechanism of deep learning improve on performance indicators,and achieved good results in the embedded platform.