Research Of Image Encryption Based On Evolution Algorithm And Cellular Automata

In recent years,with the continuous development of digital information and communication technology,the storage and transmission of digital images has become the norm.The security of digital image information has become the focus of people.Because the image has the characteristics of large amount of data and strong pixel correlation,the security and efficiency of image encryption using traditional encryption methods can not be guaranteed.This requires researchers to design algorithms for image encryption.Evolutionary algorithm has good robustness and optimization characteristics.Evolutionary algorithm using in image encryption can evolve an encrypted image with high security performance or the best key.Cellular automata is widely used in image encryption because of its rich operation rules,highly parallel structure and complex nonlinear dynamic characteristics.Chaotic systems are also widely used in the field of cryptography because of their high sensitivity to initial values and good pseudo-random characteristics.In this thesis,I studied the encryption technologies such as evolution algorithm,cellular automata and chaos,and improved two image encryption algorithms:(1)Based on the characteristics of genetic algorithm,one-dimensional cellular automata,Lorenz hyper-chaos and logistic map,an image encryption algorithm is improved.Firstly,Lorenz hyper-chaos is used to generate pseudo-random sequence,and the original image is confused according to the sequence.Then eight CA rules are selected to obtain a new pseudo-random sequence using the chaotic sequence for the diffusion step.And initial encrypted image is obtained.Then,the initial population is obtained by encrypting with logistic map,and the optimized encrypted image is obtained by genetic algorithm's evolution.(2)Based on the TLBO,two-dimensional reversible CA,Lorenz hyper-chaos and logistic map,an image encryption algorithm is improved.Firstly,Lorenz hyper-chaos is used to generate pseudo-random sequence for the confusion step.Then,the image is segmented according to bit,the high bit-image is iterated by reversible CA,and the low-bit image is XOR diffused to obtain the initial encrypted image.Take the initial encrypted image as the initial population.Then TLBO is used to optimize the best key of Logistic map to get the final encryption image.Both algorithms above use secondary encryption.Simulation show that under the condition of respectively taking the correlation coefficient and information entropy as the cost function,the correlation coefficient is closer to 0 and it can reach to 10^-6approximately.And information entropy is closer to the ideal value 8 and it can reach to7.998017 and 7.998166.And they both have large key space.They can effectively resist typical attacks such as differential attack and noise attack.They have good encryption effect and security.