Abstract:In recent years, people have put forward various image encryption algorithms based on pixel level. In fact, bit level encryption has better effect than pixel level encryption. Therefore, this paper proposes a new bit-level image encryption algorithm based on Back Propagation (BP) neural network and Gray code. Firstly, the plaintext image is conversioned into binary image, then, the hyperchaotic Lorentz system is used to generate two sets of chaotic sequences for the Gray code bit-level permutation operation to generate the permutation matrix. Secondly, the permutation matrix is converted into a bit matrix reverse order output to generate a diffusion matrix. Finally, the algorithm uses a BP neural network composed of Logistic map and Piece-Wise Linear Chaotic (PWLCM) map to generate a key stream. The key stream is xored with the diffusion matrix to generate a ciphertext matrix. The experimental results show that the algorithm improves the encryption efficiency, has good security and can resist common attack methods.

Bit-level image encryption algorithm based on BP neural network and gray code