An accurate and high-efficient QuBits steganography scheme based on hybrid neural networks

Abstract

This paper presents a novel scheme for QuBits steganography based on adaptive neural networks. Steganography based on qubits string along with the adaptive neural networks with the recycling of the modified particle swarm optimization algorithm, and using the enhanced general controlled NOT gate and NEQR representation model with the optimal target of the quantum ANNS (QANNs).In this scheme, the cover image is trained to be more accrued. Then in the obtained stego file, coefficients are classified based on their XORs. The suggested scheme avoids attacking of the sensitive data in a way that receiver can extract the information without any errors. Considering the preformed classification, secret qubits will not be revealed in the transferring process and then with the use of inverse extracting, stego file will be obtained. The most important features that our work obtained are good adaptation with human vision system and retrieval of data without getting error. Simulation results show that our proposed scheme has a good adaptation with human vision system (HVS) and outperforms in terms of PSNR factors over recently published works. Particularly, the suggested results can easily realize optimal target value locations of the ANNs to obtain low noise and high accuracy. Additionally, the proposed scheme can separately conceal a secret message and contents of the cover file.