Research On Hyperchaotic Image Encryption Based On Dynamic Filtering And Joint Permutation And Diffusion

In our daily life,hundreds of millions of images are generated every day.Image is rich in visual information.It is a very important task to protect image information security in military,medical and other fields.Therefore,image security has gradually become one of the most popular research topics in the area of information security.Image encryption is one of the important ways to ensure the image security.The application of image encryption technology can make the encrypted image have no information to the unauthorized holders.At the same time,the encrypted image can be decrypted with the right key to make the original content visible to the authorized users.At present,there are many excellent encryption algorithms in the field of cryptography,but the traditional encryption algorithms such as DES,AES and RSA are inefficient and vulnerable to attack because images have strong correlation,high redundancy and large amount of data.Chaotic system has the advantages of pseudorandomness,unpredictability,ergodicity,and extremely sensitive to initial values and parameters.It is very suitable for images with large amount of data,high redundancy and high correlation.So we need to design a new image encryption algorithm based on hyperchaotic system to ensure the security of the image.In this thesis,an image encryption scheme combining hyperchaotic system,image filtering and DNA diffusion is proposed.Firstly,the improved image filtering is used to diffuse the image and change the value of the pixels.The improved image filtering makes the filtering operation reversible,and all the filter kernels' parameters come from hyperchaotic sequence.And then the data is encoded into DNA level data.At last the DNA level data is scrambled and diffused by hyperchaotic sequences.A large number of experiments on several public images show that the proposed algorithm has a large key space,is very sensitive to the security key,has good statistical characteristics,and can resist various attacks.However,the image filtering operation uses a fixed shape kernel,which loses the diversity of filter kernel.Moreover,the proposed algorithm is not associated with plaintext data and cannot resist the chosen plaintext attack,so it needs to be further improved.Then,this thesis proposes an image encryption scheme based on 6D hyperchaotic system,joint plaintext dynamic filtering and DNA diffusion.The scheme improves the image filtering,so that the shape and weight of the filtering kernel corresponding to each pixel are not the same,and it is highly related to the hyperchaotic sequence and the original image,which is called dynamic filtering.The final results show that the scheme has high security and can effectively resist statistical analysis and differential attacks.Although the effect of dynamic filtering is very good,it is easy for attackers to break the algorithm conducts permutation and diffusion separately,which increases the risk of cracking the system.And the initial values of hyperchaotic sequence is always fixed,which also has the risk of being analyzed by a large number of statistical attacks.Therefore,in the end,this thesis proposes an image encryption scheme based on 4D chaotic system,plaintext related dynamic filtering and joint permutation and diffusion.This scheme uses SHA-256 function to calculate the hash value of plaintext image,and further generates the initials value of hyperchaotic sequence,and then apply dynamic filtering on plain image.Finally,a novel joint permutation and diffusion algorithm is proposed to complete the encryption operation.Experimental results show that the algorithm has good performance,high efficiency and can resist all kinds of attacks.