An Adaptive Chaos-based Image Encryption Scheme

An adaptive color image encryption scheme using chaotic generalized Arnold map is proposed. The three base color components of the plain-image with 256 grayscale levels are decomposed into 6 parts: 3 higher bit-planes parts and 3 lower bit-planes parts. The intensity values of all these 6 parts are summed to get one secrete key used to scramble the resulted 6 matrices by sort-based approach in a cross way. Chaotic sequences produced by one generalized Arnold map are applied to diffuse the scrambled three base color components crossly as well to get more encryption effects. An adaptive strategy is adopted in both the scrambling process and diffusion process, implying that the key streams are all related to the content of the plain-image. Therefore the image encryption scheme shows strong robustness against known/chosen plaintext attacks. Other performances are carried out, such as key sensitivity, histogram, correlation coefficients, information entropy and difference attack analysis. All the experimental results show that the proposed scheme is secure and effective for practical application.