Quantum Image Encryption Based On Daubechies D⁽⁴⁾ Quantum Wavelet Transform And Quantum Convolutional Neural Networks

The rapid development of information technology has brought people into the era of big data,but also made us must face the threat of information leakage.As a kind of common information carriers in daily life,images are easily attacked by attackers.Therefore,to ensure the secure transmission of image data,many image encryption schemes have been endlessly putting forward by related researchers.At present,people's demand on image quality is getting higher and higher,which makes the original images with large data redundancy occupy more hardware resources and also greatly reduces the efficiency of image encryption.In recent years,with the research on quantum computing gradually deepening,its parallelism just provides a solution to the problem of low efficiency on high-quality image encryption.Based on the existing quantum image encryption algorithms and some classical image encryption schemes,two quantum image encryption algorithms are proposed in this dissertation.The concrete work of this dissertation is as follows:A quantum image compression and encryption algorithm based on Daubechies D⁽⁴⁾quantum wavelet transform(DQWT)and three-dimensional hyper-chaotic Henon map is proposed.A quantum image is scrambled by an iterative generalized Arnold transform to eliminate its block effect.DQWT and the measurement matrix constructed by Hadamard gate are utilized to compress the scrambled quantum image to produce quantum compressed image.A set of fixed-length keys are generated by iterating three-dimensional hyper-chaotic Henon map,which is mapped within the pixel value range of the gray image according to the modular operation,so that it can be constructed into a quantum key image.The XOR operation is performed on the constructed quantum key image and the generated quantum compressed image to complete image encryption.Since three initial values and two control parameters are involved in the encryption process,the key space is large enough.The experimental results show the proposed quantum image encryption algorithm is feasible with acceptable resistance against some common attacks effectively.Based on the research of the existing quantum circuit model of convolutional neural network,a new model of quantum convolutional neural network model is built,which is utilized to propose a symmetric quantum image encryption algorithm.The quantum chaotic map sensitive to its initial values is employed to generate a set of keys to be utilized in the proposed quantum image encryption algorithm.To verify the security and the feasibility of the proposed scheme,three communication parties,namely sender,receiver and attacker,are assumed.The sender will inject the plaintext image and the generated key into its trained quantum neural network to encrypt in a manner similar to XOR operation and output the corresponding encrypted image.The receiver receives the encryption image and decrypts the encryption image to obtain plaintext information with keys and operations similar to those of the sender.Because the attacker does not have the key,even if the attacker is given more optimization operations,the attacker can not obtain any information of the plaintext image.Simulation results show that the quantum image encryption algorithm is secure and feasible.