Optimal quantum image encryption algorithm with the QPSO-BP neural network-based pseudo random number generator

An optimal quantum image encryption scheme named QPSO-BP-PRNG is presented by integrating back propagation (BP) neural network, quantum particle swarm optimization (QPSO) algorithm, the chaotic map-based PRNG with quantum fractional Fourier transform (QFFrT) and double random-phase encoding (DRPE) techniques to improve the effects of traditional pseudo random number generator (PRNG) for image encryption algorithms. Firstly, to achieve the exceptional incipient parameters and shorten the training time of the BP neural network, the initial weights and the threshold values of the BP neural network are renovated by the QPSO algorithm. Subsequently, the optimized QPSO-BP neural network is adopted to train the sample sequences produced from the quantum logistic map-based PRNG, which eliminates chaotic periodicity due to the strong nonlinear variability of the neural network. Lastly, the new sequences yielded with the enhanced QPSO-BP-PRNG are involved in the encoding of phase angles and the scrambling operation of the encryption procedure. According to the simulation results, the error ratio of the QPSO-BP neural network is within 0.00008 after 1000 iterations, moreover, the proposed quantum image encryption scheme performs well in security as well as reconstruction quality of decryption image.