Optical Image Encryption Technology And Cryptanalysis Based On Integral Imaging

With the in-depth development of Internet technology,the world is more closely connected,and a large amount of electronic information has appeared in people’s lives.However,information security issues also affect all aspects of human society.Information has gradually become the driving force of this era.While promoting social development,it is also important to respond to the challenges brought by information security.The application of optical theory and technology in the field of information security can make full use of the advantages of fast optical data processing,high degree of freedom,and strong robustness,and provide a high-quality optical image security system.Integral imaging technology is an imaging technology with great scientific research value and market potential,has the advantages of full parallax,true color,and no visual fatigue,so it is widely used in various fields.In the field of three-dimensional data security,integral imaging technology has received more and more attention because of its high robustness,security,simplicity,and feasibility in the calculation.We explored the application of integral imaging technology in the field of optical information security.The image encryption schemes based on integral imaging and cryptanalysis technology for integral imaging encryption algorithms are proposed.The research contents are as follows:1.Aiming at the problem of patterns in ciphertext images caused by DNA(Deoxyribonucleic acid)encoding algorithms in encryption system based on integral imaging,an image encryption algorithm based on integral imaging and cellular automata DNA encoding is proposed.The coding and decoding rules of each pixel in the element image are randomly defined by the cellular automaton.Therefore,the arrangement information of the pixel value of the element image is removed,so that the attacker cannot obtain any valid data according to the statistical characteristics of the image.which leads to the improvement of the security of the encryption system.In addition,a high-quality computational integral imaging reconstruction algorithm is applied to effectively improve the viewing quality of decrypted images.Experiments prove the feasibility and security of the proposed encryption algorithm.2.An asymmetric image encryption algorithm based on integral imaging is proposed.Since most image encryption algorithms based on integral imaging are symmetric encryption processes,they are vulnerable to chosen plaintext attacks.Therefore,we propose an asymmetric encryption algorithm based on integral imaging.Through the random vector decomposition technology based on the maximum length cellular automaton,an effective one-way trapdoor function is introduced into the encryption system.It improves the system’s ability to resist attacks.At the same time,the high randomness of the maximum length cellular automaton is used to improve the randomness of the ciphertext image.Provides a high-security integral imaging asymmetric image encryption system.3.A color image encryption method based on deep learning demosaicing and integral imaging is proposed.Different from the traditional three-channel encryption method of color three-dimensional image encryption algorithm,we use Bayer color filter array to reduce the complexity of the encryption system to realize single-channel integral imaging color image encryption.The down-sampled data is encoded by the proposed adaptive block cellular automata algorithm,which ensures the security of the algorithm.The demosaicing algorithm based on deep learning uses the correlation between pixels of different colors to further restore the original full-color image.It effectively solves the problem of low resolution of reconstructed images in traditional algorithms.It is worth noting that because the data volume of the captured downsampled image is reduced,the time cost of the algorithm is reduced compared with the previous color three-dimensional image encryption method.4.An optical asymmetric cryptanalysis model based on deep learning is proposed,which is suitable for a variety of different asymmetric encryption algorithms.Through a specific neural network model to learn the mapping relationship between ciphertext and plaintext in different asymmetric optical cryptosystems.The finally obtained optimized network model will be used as the decryption key.During the test,by using the optimized network model,the features of the plaintext image can be extracted from the corresponding ciphertext.Compared with traditional cryptanalysis methods,the proposed network model can extract unknown plaintext from a given ciphertext without using various optical keys and complex phase retrieval algorithms.5.A cryptanalysis technology for the integral imaging image encryption system is proposed,which proves the fragility of the optical security system based on integral imaging without directly obtaining the optical key.A series of ciphertext plaintext pairs obtained by the integral imaging symmetric encryption algorithm are used to train the conditional generation adversarial network.Then,through the network model,the unknown plaintext can be retrieved directly from the given ciphertext without additional conditions.In addition,the proposed network model is also effective for cryptanalysis of optical asymmetric encryption algorithms based on integral imaging.The proposed attack method based on deep learning provides a new direction for cryptanalysis of optical encryption based on integral imaging.