Secure chaotic image encryption method using random graph traversal and three step diffusion

In this paper, we have presented an image encryption technique using random graph traversal, chaotic maps, and a secure hashing algorithm. Due to the innate features of images, such as a high correlation between pixels, high redundancy, and the bulk capacity of images, traditional cryptography methods cannot be used for image encryption. Several theories can be exploited for image cryptography, but we choose random graph traversal because of its immense potential in this field. Chaotic maps are highly sensitive to initial conditions, which makes them suitable for encryption tasks. It is difficult to predict the dynamic properties of high-dimensional maps since they are more complex in nature. We use a three-dimensional Lorenz map to make our algorithm more robust. The initial parameters of the Lorenz map are modified by using the SHA-3 algorithm in the plain image. All the pixels of the image are assumed to be nodes of a graph, and a novel method is used to find a random path of pixel nodes in this graph network using a Lorenz map sequence such that each pixel node is visited exactly once. Finally, the chaotic sequences are used to diffuse the image into three sub-parts to obtain the final cipher image. The experiment has been performed on several standard images to compare the proposed approach with related methods in terms of qualitative and quantitative measures. The proposed approach fails to decrypt the cipher image even with a key deviation as low as 10(-7) compared to related methods, which show excellent key sensitivity compared to the proposed method. Experimental results show that the proposed algorithm is highly secure and can resist several attacks.