A chaos-based image encryption algorithm based on multiresolution singular value decomposition and a symmetric attractor

This paper proposes a new image cryptosystem using multiple chaotic maps and multiresolution singular value decomposition (MR-SVD). The encryption process starts with implementing the MR-SVD to decompose the original image into the four fundamental sub-bands, i.e., Approximation (A), Vertical (V), Horizontal (H), Diagonal (D) sub-bands respectively. Since the approximation part gives the most information about the image, this sub-band is selected to perform permutation and diffusion. The permutation of all the four sub-bands is done by deploying the Baker map. The diffusion of pixels in the permuted approximation part is done by our neighbourhood diffusion scheme that uses the numerical solution of the chaotic Thomas' cyclically symmetric attractor. The final cipher is obtained by combining all the four (A, H, V, D) partial ciphers by performing inverse MR-SVD. The experimental results of our proposed scheme on various benchmarks tests indicate that the algorithm is highly secure and can withstand various attacks.