Research On Video Encryption Method

With the rapid development of smart cities and the Internet,videos have been widely applied in fields such as defense,education,public safety,entertainment.A large number of videos data are transmitted on the network at every moment.While the video brings convenience to human life,security issues it brings have attracted more and more attention.Video encryption is an important measure to improve video security.Although the existing encryption algorithm can achieve good encryption effect in the text encryption,it can not achieve good results in video encryption.In order to improve the security of video encryption,this paper did some research on video encryption methods from hyperchaotic systems,Faster R-CNN(Regions with Convolutional Neural Network),and regions of interest.Firstly,this paper studied an image encryption method based on hyperchaos system and the correlation of pixel information.The method proposed in this paper added not only the plaintext image in the encryption process,and the pixel data information,and the pixel position information while the traditional image encryption method considers the less plaintext information of the image which can increase the security of encryption.The encryption method adopted a three-level structure of diffusion encryption,scrambling encryption and diffusion encryption.The encryption keys are generated dynamically by different hyperchaotic sequences and pixel information.The three-level structure of encryption can ensure the sensitivity of encryption.In the process of diffusion encryption,the position information of the plaintext pixel is added,and the data information of the plaintext pixel is added in the process of scrambling encryption.By adding different plaintext information in different encryption process,the relevance of the encrypted plaintext can be improved.The result of experiment demonstrated that the encryption method proposed in this paper has the characteristics of large key space,low correlation between adjacent pixels,large information entropy,high encryption sensitivity,and resistance to differential attack.Secondly,based on the image encryption method proposed in the previous work,the paper proposed a video encryption method based on the region of interest.It took full account of the video content information and encrypted different video contents with the different encryption algorithms of different security levels by combining the Faster R-CNN and the video encryption.First of all,the interest dataset and the Faster R-CNN are used to produce an interest model to extract the region of interest in the video,and then it is used to extract the region of interest in the video,and the video will be divided into the region of interest and the region of non interest.Finally,the image encryption algorithm proposed in the previous work with high level of security is used to encrypt the region of interest in video,and the diffusion encryption algorithm based on (17)domain is used to encrypt the region of non interest of video.After all of this,the cipher of the encrypted video is generated.