In this paper, an efficient image compression encryption scheme is proposed. Firstly, compressing the image by using BP neural network before encryption, which not only saves the subsequent encryption and decryption time but also minimizes storage and transmission space. Secondly, Chialvo neuron model is used to generate chaotic sequences for the encryption process in the scheme, confusing the compressed image according to a confusion algorithm combining single-plane and cross-plane. Next, diffusing the confused image. Subsequently, the diffused image and chaotic sequences are DNA encoded, performing DNA same or operation (DNA mutual mutations) and DNA mutations (DNA self-mutations). Finally, the DNA mutations results are decoded to obtain the cipher image. Security tests and compression performance analysis show that the scheme can efficiently and securely compress and encrypt images of different sizes with higher reconstruction quality compared to other compression schemes.
To effectively deal with the complex and changeable network environment and meet the high standard of information transmission security, this paper designs a convenient and efficient image encryption algorithm on the incommensurate fractional-order neural network model system. The algorithm uses the generated random number sequence to construct an index matrix and combines the Hilbert curve traversal principle to displace each position of image pixels. The image is converted into a DNA-encoded image by DNA coding technology. The DNA addition principle completes the diffusion operation, and the ciphertext image is obtained. Simulation results and security tests show that the proposed algorithm can encrypt image information effectively and has good anti-attack capability.
27 October 2025 in Journal of King Saud University Computer and Information Sciences
To tackle the issue of efficiency and security in the multi-image encryption scheme, a scheme is designed. Firstly, the combination of the keystreams of hyperchaotic map and fractional-order Lorenz system are selected for the scheme. Then, multiple images are reshaped into a vector and the vector is reshaped into a cube. A first type of diffusion is performed on the cube and this procedure is based on mixed arithmetic operations. The diffusion equation is characterized by both a plaintext delay term and a ciphertext feedback term. Secondly, the cube is then confused in a point-to-point manner by the scheme, and the large cube is segmented into four or eight small cubes. The segmentation schemes are determined by plaintext parameter. The rotation of small cube is selected according to the different schemes, and a point-to-point confusion is used to confuse the small cubes that are not rotating. The confusion procedure of small cube is similar to the first confusion step. Then, a combination method corresponding to the cube segmentation method is used to combine the small cubes into a large cipher cube, which is similar to the game of segmenting and assembling building block. Thirdly, the second type of diffusion has different mixed arithmetic expressions. The keystream values are changed according to the parameter values of iteration during the confusion procedure. The relative positions of the keystreams are made variable to increase the difficulty of brute-force cracking. Simulation results and security analysis demonstrate that images can be efficiently and securely encrypted.
In continuous neural modeling, memristor coupling has been investigated widely. Yet, there is little research on discrete neural networks in the field. Discrete models with synaptic crosstalk are even less common. In this paper, two locally active discrete memristors are used to couple two discrete Aihara neurons to form a map called DMCAN. Then, the synapse is modeled using a discrete memristor and the DMCAN map with crosstalk is constructed. The DMCAN map is investigated using phase diagram, chaotic sequence, Lyapunov exponent spectrum (LEs) and bifurcation diagrams (BD). Its rich and complex dynamical behavior, which includes attractor coexistence, state transfer, Feigenbaum trees, and complexity, is systematically analyzed. In addition, the DMCAN map is implemented in hardware on a DSP platform. Numerical simulations are further validated for correctness. Numerical and experimental findings show that the synaptic connections of neurons can be modeled by discrete memristor coupling which leads to the construction of more complicated discrete neural networks.