Image Encryption Based on Hopfield Neural Network and Bidirectional Flipping

Many encryption systems face two problems: the key has nothing to do with the plaintext; only a single chaotic sequence is adopted during the encryption. To solve the problems, this paper proposes an image encryption method based on Hopfield neural network and bidirectional flipping. Firstly, the plaintext image was segmented into blocks, the resulting image matrix was block scrambled, and each block was bidirectionally flipped to complete the scrambling process. After that, the plaintext image was processed by the hash algorithm to obtain the initial values and control parameters of the chaotic system, producing a pseudo-random sequence. Then, a diffusion matrix was generated through the optimization by Hopfield neural network and used to derive a ciphertext image through diffusion transformation. Experimental results show that our algorithm is highly sensitive to plaintext, strongly resistant to common attacks, and very efficient in encryption.