Research On Application Of Cellular Neural Networks To Image Encryption

In recent years, the rapid development of computer and network technologyprovides a broad application platform for multimedia information such as audio andvideo data. Multimedia information is used more and more widely for its imageintuitive and vivid. However，there exist potential safety problems of multimediainformation while it brings convenience to our life. Cryptography, as a key technologyto guarantee transmission data security, is facing new challenges in the multimedia datatransmission.Contrasting to common text data, multimedia information usually containsvolumes of data, which make encryption run very slowly and consume huge amount ofpower when we encrypt it with traditional cryptosystem. The research accomplishmentsof digital image, which is a main form of multimedia data, can be easily applied to othermultimedia data. During recent years, the image processing technology based onCellular Neural Network (CNN for short) has been widely studied, especially forCellular Neural Network with chaotic characteristic and parallel processing capability.Both the features make Cellular Neural Network very suitable for image encryption.This paper focuses on analyzing and designing novel image encryption schemes basedon Cellular Neural Network. The main contents of this paper are listed as follows.①The origin and development of neural networks and Cellular Neural Networks areintroduced. First, this chapter gives a general introduction, including the origin,development and the main research results in the field of neural networks; then, itintroduces the Cellular Neural Networks based on the theories of the Hopfield neuralnetworks and cellular automata, and analyzes the advantages Cellular Neural Networkinherits from both the neural networks and its own characteristics.②A detailed introduction of the basic knowledge of Cellular Neural Networks anddigital image encryption are introduced. First, the mathematical model, networkstructure and dynamic performance of the Cellular Neural Networks are introduced; andthen, this chapter gives a brief introduction of the basic knowledge of cryptography, theresearch background of digital image encryption algorithm and several major imageencryption algorithms; finally, gives some evaluation indexes of encryption algorithm.③A novel image encryption algorithm is proposed based on Cellular NeuralNetwork. The main objective of this algorithm is to solve the problem of traditional stream cipher’s insensitivity to the change of plain text by using a hyper chaotic systemof6-D CNN as the key source, selecting the secret key based on the results of logicaloperations of pixel values in the plain image, and introducing simultaneously bothposition permutation and value transformation. It is shown that both NPCR values andthe sensitivity to key (>0.996) can meet the security requirement of image encryption.The simulation results also indicate that the algorithm is relatively easy to realize withlow computation complexity, and assures, accordingly, the secure transmission of digitalimages.④A framework of block image encryption cipher based on CNN is proposed toimprove the efficiency and security, where the properties of parallel computing andlocal diffusion of CNN are utilized effectively. An algorithm based on feistel is alsogiven under this framework. In each round, a cell encrypts one block image data withblock encryption algorithm first, and then effected by the output of8-neighbouring cells.The algorithm is suitable for VLSI implementation as well as parallel computing. Boththe analyses and simulations prove the high security of the proposed algorithm.