A new image encryption algorithm is proposed by combining fractional Hermite transform (FHT) with hyper-chaos map. In the first round of pixel scrambling, the original image is performed by the Zigzag operation and the Arnold transform. Then the scrambled image matrix is XORed with an encryption sequence composed of an 8-bit linear feedback shift register. Compared with the traditional Hermite transform, the FHT has better noise tolerance, and it could process the non-stationary input signals better. Subsequently, the encrypted data is re-encrypted by the cyclic shift operation based on the hyper-chaotic system. The four initial values of the hyper-chaotic system are obtained by combining plaintext image with SHA-512. Finally, the processed image matrix is XORed with a chaotic matrix generated by the three-dimensional Logistic-sine-coupling map. The presented image encryption algorithm has a large key space and strong anti-noise ability. Simulation results show that the proposed algorithm is feasible, secure and effective.

提出了一种新的图像加密算法，将分数Hermite变换(FHT)与超混沌图相结合。在第一轮像素加扰中，原始图像通过Zigzag操作和Arnold变换进行。然后将加扰的图像矩阵与一个由 8 位线性反馈移位寄存器组成的加密序列进行异或。与传统的 Hermite 变换相比，FHT 具有更好的噪声容限，能够更好地处理非平稳输入信号。随后，加密数据通过基于超混沌系统的循环移位操作重新加密。超混沌系统的四个初始值是通过将明文图像与 SHA-512 相结合得到的。最后，将处理后的图像矩阵与由三维 Logistic-sine-coupling 映射生成的混沌矩阵进行异或。提出的图像加密算法密钥空间大，抗噪能力强。仿真结果表明，该算法可行、安全、有效。