ABSTRACT 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.

中文翻译：

---

一种基于多分辨率奇异值分解和对称吸引子的混沌图像加密算法

摘要 本文提出了一种使用多混沌映射和多分辨率奇异值分解（MR-SVD）的新图像密码系统。加密过程从实施MR-SVD 开始，将原始图像分解为四个基本子带，即分别为近似(A)、垂直(V)、水平(H)、对角线(D) 子带。由于近似部分提供了有关图像的最多信息，因此选择该子带进行排列和扩散。所有四个子带的排列都是通过部署 Baker 地图来完成的。置换近似部分中像素的扩散是通过我们的邻域扩散方案完成的，该方案使用了混沌 Thomas 循环对称吸引子的数值解。最终的密码是通过组合所有四个（A、H、V、D) 通过执行逆 MR-SVD 的部分密码。我们提出的方案在各种基准测试中的实验结果表明，该算法是高度安全的，可以抵御各种攻击。