Deep Neural Network-based Cryptanalysis Of Computational Ghost Imaging Encryption System

As the degree of information continues to increase,how to ensure the security of information has become increasingly severe.Optical information security technology has achieved rapid development in the field of information security due to its parallel processing and multi-dimensional design.Among them,computational ghost imaging as a new type of optical imaging technology,its natural non-locality and data compression characteristics are widely concerned by researchers related to optical information security.However,any encryption technology requires a rigorous security analysis to verify that its encryption system is truly stable and reliable.This paper firstly studies the research background and development status of computational ghost imaging technology.At the same time,the basic concepts of optical information security,scalar diffraction and basic theory of neural network are introduced one by one.Furthermore,the imaging mechanism of computational ghost imaging and the performance of its encryption system are further studied,and the security evaluation based on computational ghost imaging encryption system is carried out through deep neural network.The main work of the thesis includes the following two aspects:(1)The principle of ghost imaging,computational ghost imaging and optical encryption technology based on computational ghost imaging is introduced.The imaging mechanism of computational ghost imaging is verified by experiments.The nonlinear response problem in computational ghost imaging experiments is studied,which improves the imaging quality of computational ghost imaging in practical experiments.At the same time,the process of encryption and decryption of ghost imaging is studied by numerical simulation,and the security performance based on computational ghost imaging encryption system is evaluated by three corresponding attack methods.(2)Further,using the deep neural network,a method for calculating the security analysis of the ghost imaging encryption system is proposed.By mapping a series of known "ciphertext-plain text pairs" into the designed deep neural network structure,the mapping relationship between the ciphertext and the plaintext of the system is obtained,that is,the equivalent of the system.Key"(can be considered as "equivalent decryption network").The effectiveness and robustness of the proposed method are verified by numerical simulation.Finally,using the attack method proposed in this paper,an enhanced encryption scheme based on computational ghost imaging is tested.It is found that the proposed attack method can still successfully obtain the corresponding "equivalent key",and the above traditional attack method No.