Abstract
Based on elementary cellular automata, a new image encryption algorithm is proposed in this paper. In this algorithm, a special kind of periodic boundary cellular automata with unity attractors is used. From the viewpoint of security, the number of cellular automata attractor states are changed with respect to the encrypted image, and different key streams are used to encrypt different plain images. The cellular neural network with chaotic properties is used as the generator of a pseudo-random key stream. Theoretical analysis and experimental results have both confirmed that the proposed algorithm possesses high security level and good performances against differential and statistical attacks. The comparison with other existing schemes is given, which shows the superiority of the proposal scheme.