Research On Image Encryption Using Chaos Theory

Digital image, which is a significant aspect of multimedia information, has played an important role in peoples’daily life. However, there are a lot of threatens during the digital images saved or transported due to their intrinsic features. The security problems have become the main concerns in source coding, channel coding, computer technology and mathematics fields, and they are the urgent problems.Cryptography technologies are most common used and efficient methods to protect the digital information, which have been widely applied to solve most of the security issues. However, the recent researches illustrate that the traditional cryptosystems, such as DES and AES, are not suitable for encrypting multimedia information due to their intrinsic features. The development of chaos theory and the natural similarities between the chaos theory and cryptography make chaos-based cryptosystems good candidates and provide a new method for encrypting multimedia information. From 1980s, chaos-based cryptography has developed greatly. In this paper, we research and summarize the recent achievements of chaos-based cryptosystems, and further propose two kinds of chaos-based image encryption architectures in pixel-level and bit-level, respectively. In each kind of architecture,2 or 3 cryptosystems are proposed in different perspectives to solve distinct problems. Simulations have been carried out to guarantee the security and efficiency of proposed cryptosystems.The innovations of the paper are summarized below:(1) The security problems of traditional architecture of chaos based image encryption algorithms are analyzed, and we propose a new compound mapping based on ordinary and reverse 2 dimensional chaotic map, in which a new mapping rule that from a pseudorandom position to anther pseudorandom position is proposed based on the distinct features of ordinary and reverse 2 dimensional chaotic maps. A new chaos-based image encryption algorithm based on reverse 2 dimensional chaotic map is proposed. Compared with the ordinary 2 dimensional chaotic map, the new mapping rule is more complex and random, and can be resisted to chosen plain-text attack.(2) The independent problems of confusion and diffusion are analyzed, and a new dependent diffusion architecture is proposed, in which confusion and diffusion operations are fully mixed up rather than treats them as two separated steps, which make it resistant to known/chosen plaintext attack. A new image encryption algorithm based on the new architecture is proposed to prove the high security and efficient of new architecture.(3) Considering the features of digital images, the problem of high correlation of adjacent pixels in permuted image is analyzed, and "divide and rule" strategy is used and a new kind of correlation separation mapping and correlation recovery mapping are proposed and applied in chaos-based image encryption algorithms. Furthermore, a novel random area permutation and random rotation are proposed to enhance the permutation effect after confusion phase.(4) Considering the bit level features of digital images, the bit-level permutation architecture is firstly proposed. The different contributions to the pixel value by different bit planes are analyzed, and propose a new kind of bit-level image encryption, in which different bit planes are permuted independently. Two strategies are employed to permute higher 4 bit planes and lower 4 bit plane according to their different contributions to the pixels’value. The new confusion operations can not only change pixels’locations but also modify their values. Simulations have been carried out to prove the high security and efficiency of proposed scheme.(5) Two intrinsic features of digital image in bit level are proposed firstly. They are the correlation between higher bit planes are always high; the distribution of bit values are stable in macroscopic view and fluctuated in microscopic view. The three targets of bit-level confusion operation are analyzed, and propose a new kind of permutation plane, in which each element contains different value weight. A new cryptosystem is proposed based on the new permutation plane for RGB color image, which breaks the limitation of the bits movements from one bit plane to another one. In the new algorithm all the limitation of bit mapping are removed. The effect of new confusion is better.(6) The calculation redundancy of traditional chaos-based image encryption is analyzed, and we propose a new light weight confusion algorithm to reduce the redundancy. A cascade cross circle diffusion strategy is employ to solve the "stop point mechanism" in traditional chaos-based image encryption algorithms. Simulation results show that the new image encryption algorithm can achieve both high security and efficiency.