Study On Chaotic Image Encryption Algorithm Based On Hopfield Neural Network And Compressive Sensing

Chaos has the characteristics of pseudo-random and non-linear,it is sensitive to parameters and initial values.In recent years,chaotic systems have been widely used in image encryption systems.The combination not only makes the field of image encryption algorithms have a profounder academic background,but also greatly improves the security of the image encryption system.However,some image encryption schemes have drawbacks such as plaintext or ciphertext is uncorrelated,scrambling and diffusion processes are independent and so on.These image encryption schemes are vulnerable to be attacked and deciphered.In order to solve these problems,this thesis introduces two image encryption algorithms.The main work and innovations of this thesis are as follows:(1)On the basis of chaos and neural network,the scrambling and diffusion of the encryption algorithm are carried out at the same time,which overcomes the problem of scrambling and diffusion are independent in traditional encryption algorithms.Firstly,the first diffusion is performed simultaneously with the first scrambling;then the Hopfield chaotic neural network is used to generate the second diffusion matrix,and the initial values of the Hopfield chaotic neural network system are related to the plaintext,which not only can resist the selected plaintext attacks,but also improves the sensitivity of the secret keys;finally,a second scrambling is performed on some special pixels to further improve the security.Experimental results show that the image encryption algorithm in this thesis has a good security performance and can resist various attacks.(2)On the basis of chaos and compressive sensing,the image is encrypted by dynamically transforming the ciphertext image.Firstly,we use compressive sensing technology to compress and combine the plaintext images,then use chaotic sequences to scramble the pixels;finally,use chaotic sequences and noise-like random image to generate the key stream and diffusion matrix to diffuse the pixels.At last,generate a dynamic ciphertext image.The algorithm compresses the plaintext image through compressive sensing technology,which not only reduces the correlation between pixels,but also reduces the time complexity of the system;when the pixels are diffused,a part of the ciphertext image is used to encrypt another part of the plaintext which makes the encryption algorithm more confidential and safer;using noise-like random images to generate the diffusion matrix,even if the same secret keys to encrypt the same plaintext image,the ciphertext image of each encryption is different.Therefore,the proposed encryption system can resist selected plaintext attacks.Finally,experimental results verify the reliability and security of the encryption system.