Mixed gray-color images encryption algorithm based on a memristor chaotic system and 2D compression sensing

Abstract

Multi-image encryption has attracted great interest in the era of secret communication, however, the mixed encryption of gray and color images has been reported rarely. Meanwhile, the security of image encryption based on chaos system depends heavily on the performance of pseudo randomness and sensibility. Thus, a four-dimensional chaotic system is improved by introducing a memristor, which has good nonlinear characteristic and can enhance chaotic property. It is found that this new system possesses larger chaotic scope and exhibits a coexistence phenomenon, all of which reveal the intricate nature and underlying application. After the verification of spectral entropy and circuit simulations of the new system, a mixed-image encryption algorithm based on two-dimension compression sensing (2DCS) is designed innovatively. To achieve the target of compression and encryption, the image set is first preprocessed and varied utilizing DWT, and then is sampled by measurement matrix in two-dimension. Next, the pixels of compressed image set are diffused by employing DNA coding. The accuracy of reconstruction image is closely related to the randomness and sparsity of the measurement matrix. Chaotic system with pseudo-randomness can be fully exploited to fabric a measurement matrix. Hence the Toeplitz cyclic chaos matrix is conceived and the reconstruction accuracy does have better results. In addition, this method can not only encrypt multiple mixed images with grey and color simultaneously, but also compress them with different sizes and less storage space. At last, the reliability of the novel algorithm is assessed through a series of reconstruction and security performance tests. The results indicate that the encryption scheme has a better effect than encrypting one kind of image, and prove that the algorithm has high reconstruction accuracy and security.