Image encryption algorithm based on DNA encoding and CNN

In this study, a novel, more stable, secure, and reliable image encryption model has been introduced. It combines a Convolutional Neural Network (ConvNet/CNN) model with an intertwining logistic map to generate secret keys. Additionally, initial conditions, control parameters, and secret keys are employed by the intertwining logistic map to produce diverse chaotic sequences. Permutation, DNA encoding, diffusion, and bit reversion operations are applied for scrambling and manipulating image pixels. The proposed encryption model was thoroughly examined using various analysis methods such as cropping attack, histogram analysis, key space evaluation, noise attack, information entropy assessment, differential attack, key sensitivity, and correlation coefficient examination. To expand the keyspace and enhance confusion and diffusion in the proposed encryption algorithm, the model employs different subkeys, private keys, and public keys through the Convolutional Neural Network. Furthermore, numerical and perceptual results were compared with the state-of-the-art outcomes to validate the model. Ultimately, the derived results demonstrate that the proposed intertwining logistic map-based image encryption model utilizing Convolutional Neural Network outperforms existing methods. This is due to its significant improvement in information entropy, enhanced randomness, high resistance against differential and statistical attacks, and overall efficiency.