Enhanced Multi Chaotic Systems Based Pixel Shuffling for Encryption

Abstract

The problem of secure transmission of digital media is addressed and dealt using the chaotic systems. The system of chaotic maps are considered a very interesting complex nonlinear phenomenon in the last four decades. Encryption of data is the most prevalent technique used for the transmission of highly sensitive information. This is achieved by using the chaotic maps for shuffling the image bits in order to increase randomness and confusion among the pixels of image. We have performed color image encryption using the chaotic maps by exploiting its intrinsic features. A set of four chaotic maps namely Hénon map(discrete time), Lorenz (butterfly attractor), Chua (double scroll attractor) and Rössler (spiral attractor) are used and the chaotic sequences generated by them acts as the key to the cryptosystem. We have proposed a new shuffling pattern of plaintext image bits using the information contained in the image and design a scheme for the resistance to two differntial attacks, chosen-ciphertext attack(CCA) and known-plaintext attack(KPA). The method first preprocesses the plaintext color image to convert it into binarized form, and then evaluate the number of binary 1s in the image. Now, the encryption scheme makes use of this information to generate the chaotic sequences. A completely different set of sequences is generated for different plaintext images. The indices of the chaotic sequences are used for the encryption of RGB image by shuffling the bits of the three components columnwise and then rowise in a pair of two bits. The dependence of chaotic sequences on the original image makes the scheme resistant to the attacks. Thus, by analyzing any plain/cipher image pairs one cannot deduce the key of the cryptosystem.In order to perform the statistical measure, Number of Pixel Change Rate (NPCR), UACI (Unified Average Changed Intensity) , histogram analysis, information entropy and correlation coefficient are used as parameters to evaluate the performance of the proposed method. It is also shown that the ciphered image is very sensitive to a slight change in the bit values of original image. Eventually, empirical images and results are used to prove the higher security and performance of the proposed approach over the existing algorithm.