Chaotic Encryption Algorithm And Optimization Based On Face Detection Of Digital Image

The Cloud computing,Internet of things,artificial intelligence,block chain and other technological elements and data elements are interrelated and deeply integrated,which is the inevitable trend of the development of digital industry,and this interconnection has been verified.In this interconnected relationship,information security is becoming more and more important,and the corresponding requirements for the protection of user privacy will continue to be postponed,although there are laws to protect people's privacy.However,general privacy disclosure incidents will invest a lot of public resources if legal measures are taken,so it can be seen that it is the trend of the times to take feasible technical measures to protect privacy.People store and share images on the Internet platform,and the cameras of the Internet of things monitor and capture natural scenes in daily life.These data often contain users' personal privacy and biometric information.More personal information can be gained by mastering face images,which is of great significance to the protection of face image information.With the rapid development of deep learning technology,the neural network is used to detect the face of the image,and the face region is selected for encryption.Compared with the full encryption of the image,the face region of the image is detected by deep learning,and selecting the face region for encryption can decrease the amount of computation,save time and cost,and meet the needs of users to protect personal privacy.In this paper,we design an image chaotic encryption algorithm based on face detection.The main research contents are as follows:(1)Face detection MTCNN algorithm has the characteristics of fast face detection speed and small space occupancy.In view of its simplified structure and large space for improvement,this paper first optimizes the MTCNN.After optimizing the network structure and parameters,the trained model is used to detect the face of the image,which improves the performance of the algorithm.(2)Firstly,the pixel value of the digital image decomposes into bit planes,and the pseudo-random matrix is generated by the hyperchaotic Lorenz system.The five bit-planes of the high position are selected for scrambling encryption of the plaintext correlation,and 5 encrypted images are reconstructed into an intermediate cipher image.Finally,the diffusion strategy based on the GF17 domain is used to diffuse the intermediate cipher image to get the final cipher image.Through the correlation analysis and security analysis of the cipher image,the purpose is to test the encryption effect of the algorithm.Experimental results show that encryption and decryption of the algorithm includes bit planes decomposition and reconstruction,which makes encryption and decryption process more complex,the algorithm is very sensitive to the key,has strong ability to resist plaintext attacks,and has more reliable security.(3)Combining the above two methods,firstly,the face detection image is obtained to get the face region,then the hyperchaotic system is used to generate pseudo-random sequence,and finally the 5-bit high-bit image of the face region is selected for encryption.Encrypting the image face information stored in the cloud platform in this way can not only protect the user's face information,but also reduce the computing cost and improve the security of the encrypted image.