Chaotic Image Encryption Algorithm and Neural Network for Information Security Assurance of Media Distribution

ABSTRACT With the acceleration of digitization and globalization of media distribution, traditional encryption algorithms have problems such as high computational complexity and insufficient real‐time performance. To meet the security needs of massive multimedia data transmission and storage, this research is oriented towards information security assurance for media distribution, and conducts an in‐depth study on chaotic encryption algorithm and neural network. In this study, a “stream‐block coordination” mechanism is introduced to the logistic chaos mapping based encryption algorithm to balance the encryption efficiency and security. The Hopfield neural network is combined with the multilayer diffusion algorithm to perform nonlinear obfuscation of the data value domain. The fractional order 5‐dimensional cellular neural network introduces the Fisher Yates dislocation diffusion technique to effectively resist the threats of statistical analysis attacks in media information dissemination. The experimental results indicated that the area of the false positive rate–recall curve of the research method was about 97.8%. In the practical application test, the difference between training accuracy and validation accuracy of the research method was 0.1%. The decryption accuracy in the face of noise attack was 97.8%. The information entropy difference with the original image was 0.99. In conclusion, the research on chaotic encryption algorithm and neural network oriented towards information security assurance for media distribution has better high‐dimensional feature detectability, robustness, and generalization ability.