الفهرس | Only 14 pages are availabe for public view |
Abstract This thesis consists of eight chapters. In chapter one, we introduce an overview of the thesis and its organization. In chapter two, the main definitions and also, explain the different methods used in image encryption. In chapter three, We gave cryptanalysis of an image encryption scheme, based on CNN and chaotic logistic map. We analyzed the security of the scheme and we have shown that it can be broken with known/chosen plain-image and it is insensitive to the change of the secret plain-image. Also We gave cryptanalysis of an image encryption scheme based on elementary cellular automata. We pointed out there is a simple condition makes a number of secret keys invalid. The scheme is insensitive to the change of the plain-image. The scheme is not secured to the following three different classical types of attacks: chosen plaintext, chosen ciphertext, and known plaintext. In the three attacks, only a pair of (plaintext/ciphertext) is needed to break the image encryption scheme. We proposed a one time key method for solving these problems and defeat cryptanalysis. In chapter four, A new feedback Confusion/diffusion architecture encryption scheme based on a 3’rd-order cellular neural network and affine transformation is proposed. It utilizes a three order cellular neural network (CNN) as a generator of pseudo-random key stream sequence. Furthermore, a novel affine transformation in permutation stage is proposed. In order to produce good avalanche property and good sensitivity form cipher to plaintext, the confusion and diffusion processes are controlled by rounds times’ key. The experimental results show that the proposed encryption scheme appears a good resistance to brute force attacks, known plaintext, and ciphertext attacks. Furthermore, the random-like nature of cellular neural network is effectively spread into encrypted images. Simulation results and analysis prove the validity and safety of the proposed algorithm. In chapter five, Based on elementary cellular automata, a new image encryption algorithm is proposed, where a special kind of periodic boundary cellular automata with unity attractors is used. The rule states used in the encryption process are different from the states used in the decryption one. The key stream is generated using the cellular neural network, and the control parameters are not fixed, but determined by the plain-image. Thus, different plain images result in nonidentical control parameters and distinct key streams. Also, the secret CA rules and states are determined by the control parameters, and thus the rules and states also change with different images. The security analysis for the proposed scheme has been presented. Simulation results on some gray level images show that the proposed image encryption scheme has perfect information concealing, and satisfies the properties of confusion and diffusion. And, some other comparative results are also given to show that the proposed scheme’s security is significantly enhanced, e.g., it can resist some known-plaintext and chosen-plaintext attacks effectively, and the scheme is also more efficient in implementation, e.g., compared with AES.In chapter six, A self-adaptive image encryption based on Memory Reversible Cellular Automata (MRCA). The self-adaptive encryption is realized by using one half of image data to encrypt the other half of the image mutually. We utilized cellular neural network system as a generator of a pseudo-random key stream sequence. Simulation results on some gray level images confirm that the proposed algorithm can be realized easily while achieving high security level, high sensitivity and other randomness properties. In chapter seven, A new cellular automata rule, we call ”Sum-rule”, is use for generating random numbers. Each generated binary sequence is checked by the poker test. A designated five octal digits are treated as a poker hand with new computations. Poker test is applied to Sum-rule, and on the proposed sum rule with two demission cellular automata (2-DCA). Statistical tests show that the sum rule passes a maximum number of NIST and FIPS tests compared to other PRNG functions, where other tests not passed. When applying the sum rule with 2-DCA, the generated sequence passed all statistical test suite for random and pseudorandom number generator for cryptographic applications (NIST). . In chapter eight, Conclusion and future work are presented. |