This paper proposes a color image encryption algorithm based on a cloud model Fibonacci chaotic system, as well as a matrix convolution operation that can protect image content effectively and safely. The algorithm combines the cloud model with the generalized Fibonacci, creating a new complex chaotic system that realizes the dynamic random variation of chaotic sequences. The chaotic sequence is used to scramble the pixel coordinates of the mosaic images of the R, G, and B components of the color image. Then, the chaotic sequence value is used as a matrix convolution cloud algorithm that alternately updates the input value of the matrix convolution operation and the pixel value to obtain the permutation transformation of the original pixel value. Finally, the pixel values of the replacement and cloud model Fibonacci chaotic sequence and the pixel values of the front (rear) adjacent pixel points are subjected to a two-way exclusive XOR operation. Realizing the change of the arbitrary pixel value causes a chain transformation of the pixel values of all of the pixel points, and sequentially generates an encrypted image. Experiments show that the histogram of the encrypted image is smoother and adjacent pixels of the image have low correlation. In addition, this algorithm can resist attack experiments such as differential attack, select plaintext attack and noise attack and provides high encryption security, high anti-interference, and strong robustness. The dynamic chaotic system is used to realize the color image encryption of the dynamic key, and the encryption algorithm has higher security and the validity of the algorithm.

Aiming at the problem that the existing bit scrambling encryption algorithm is not sensitive to the bit scrambling between 8 bits of one pixel, and the anti-noise and anti-selective plaintext attack ability is weak, a 3D cyclic shift bit scrambling image encryption algorithm is proposed in this paper. Discarding the scrambling mode of 8 bits of one pixel and the bits of all pixels, the pixel values are converted into binary arrays and then converted into 3D matrices in this paper. And the higher bit-planes and the lower bit-planes, which contain the plaintext information of different weights, are scrambled by cyclic shifting respectively. Therefore the sensitivity of the bit scrambling, the anti-noise attack ability of the algorithm and the randomness of intermediate ciphertext are improved. Moreover the randomness and the anti-noise ability can be adjusted by changing the number of higher bit-planes according to different encryption requirements. A new type of Logistic-Fibonacci (L-F) cascade chaos is constructed to generate random sequences, which solves the problem of blank windows in the uneven distribution of Logistic chaos. The initial value and the control parameters are increased, the sequence randomness is improved, and the fastness of low-dimensional chaos is preserved. By strongly correlating the key with SHA-256 of the plaintext, the key stream can change adaptively with the plaintext, which greatly improves the sensitivity of the plaintext and the ability of resisting the selective plaintext and ciphertext attack. The experiments show that the algorithm can encrypt all kinds of images with high efficiency, and can resist common attacks. It is a secure and reliable image encryption algorithm.

This paper proposes a meaningful ciphertext encryption algorithm of good practicability and strong robustness. The plaintext undergoes pre‐encryption of bit scrambling and secondary encryption of discrete wavelet transform (DWT) to obtain meaningful ciphertext. In the pre‐encryption stage, only the higher three bit pages that are not all 0 or 1 are scrambled. The embedding method in the secondary encryption process is to add the hundreds and tens’ digit and ones’ digit of the middle ciphertext to the scrambled detail information obtained by DWT of the cover image. The information entropy of the intermediate ciphertext obtained by this method can reach 7.997, the correlation coefficient between the ciphertext and the cover image can reach 0.99, and the decrypted image of the ciphertext under high‐intensity pollution can still be identified. It shows that the pre‐encryption can conceal the plaintext information well, and the secondary encryption can eliminate the ciphertext texture characteristics while retaining the good robustness. Finally, the improved three one‐dimensional chaos are used to generate random sequences, and the plaintext hash value is associated with the chaotic key. The improved chaos greatly improves the randomness of the sequence and broadens the range of parameters. The hash value SHA‐256 of the plaintext is added to make the key stream adaptively change with the plaintext, which greatly improves the sensitivity of the plaintext and the ability to resist selected plaintext and ciphertext attacks. Experiments have also verified that the encryption algorithm can encrypt images of different types and shapes and has strong practicability, good bit scrambling effect, and high sensitivity to scrambling.

As a highly recognizable biometric face recognition technology, it has been widely used in many identity verification systems. In order to enhance the protection of personal privacy and ensure the safe transmission and sharing of sensitive information without affecting the user experience, this paper proposes an innovative coverless steganography framework for face recognition images based on diffusion model. The framework firstly extracts face features and generates masks containing these features. Then, combined with conditional diffusion model and text key, a deterministic Denoising Diffusion Implicit Model (DDIM) is used to sample coverless steganography images. Secret images can also be recovered in high quality with DDIM Inversion technology. A large number of experiments show that compared with the existing methods, this approach has markedly enhanced the quality of steganographic and restored images. The face recognition rate of the restored image is more than 96%, which can effectively replace the original image for face recognition. The detection accuracy of this method is 55.25% on the steganographic detection tool, which is closer to random guessing and can resist steganographic analysis. It ensures the higher security of hidden images and solves the limitation of existing methods in protecting the privacy of face images. Moreover, it is shown how to achieve controlled local steganography with a custom mask, which enhances the controllability and flexibility of the method. In conclusion, the proposed method outperforms traditional steganography in security, controllability and robustness, and provides an effective technical scheme for steganography protection of face recognition images without additional training.

Many image compression and encryption algorithms based on traditional compressed sensing and chaotic systems are time-consuming, have low reconstruction quality, and are suitable only for grayscale images. In this paper, we propose a general image compression encryption algorithm based on a deep learning compressed sensing and compound chaotic system, which is suitable for grayscale images and RGB format color images. Color images can be directly compressed and encrypted, but grayscale images need copying from 1 channel to 3 channels. First, the original image is divided into multiple 3 × 33 × 33 non-overlapping image blocks and the bilinear interpolation Bilinear and convolutional neural network are used to compress the image, so that the compression network has no restriction on the sampling rate and can obtain high-quality compression of image. Then a composite chaotic system composed of a two-dimensional cloud model and Logistic is used to encrypt and decrypt the compressed image (sliding scrambling and vector decomposition), and finally the decrypted image is reconstructed. In the reconstruction network, the convolutional neural network and bilinear interpolation Bilinear are mainly responsible for reconstructing the contour structure information, and the fully connected layer is mainly responsible for reconstructing and combining the color information to reconstruct a high-quality image. For grayscale images, we also need to calculate the average value of the corresponding positions of the 3 channels of the reconstructed image, and change the 3 channels into 1 channel. The experimental results show that the general image encryption algorithm based on deep learning compressed sensing and compound chaos system has great advantages in data processing quality and computational complexity. Although in the network the color images are used for training, the quality of grayscale image reconstruction is still better than that of other algorithms. The image encryption algorithm has a large enough key space and associates the plaintext hash value with the key, which realizes the encryption effect of one image corresponding to one key, thus being able to effectively resist brute force attacks and selective plaintext attacks. Compared with it in the comparison literature, the correlation coefficient is close to an ideal value, and the information entropy and the clear text sensitivity index are also within a critical range, which enhances the confidentiality of the image.