Research On Phase Recovery Algorithm And Its Application In Optical Cryptography

The rapid development of electronic information and communication technology has brought human into the global information age. However, people enjoy the convenience brought by development of information technology, while also bear the risk of information security issues that should not be ignored. There exist security vulnerability objectively due to the openness and own shortcomings of information system, which leads to the emergence of information encryption, hiding technology, and identity authentication technology for information interactive body. Compared with traditional information security technology, optical information security as a new generation of information security technology is just beginning to develop in recent years internationally. This thesis will deeply study phase retrieval algorithm, which is an important technique of optical information security, and explore the field of optical cryptanalysis in optical information security.The thesis briefly introduces the development of optical information security, summarizes the research results in the fields of optical cryptography and cryptanalysis. Then the basic concepts and theoretical basis of the optical information security were necessary elaborated, including Fresnel diffraction, lens properties of Fourier transform and phase retrieval algorithms. Since the phase retrieval is not merely an important method of optical information processing, it is also a significant technical means of optical cryptanalysis, the thesis carried out a detailed study in the mechanism of phase retrieval algorithm, and a hybrid iterative phase retrieval algorithm was proposed. Finally, according to the existing vulnerability of the double random phase encoding encryption system, a scheme of ciphertext-only attack was proposed. The main contributions of the thesis are described as following two points:(1) The mathematical principles of iterative phase retrieval algorithm were expatiated, and a hybrid iterative phase retrieval algorithms was proposed. Phase retrieval algorithm is a kind of inverse problem, its basic idea is to to find the intersection of two constraint sets by using optimization methods. Therefore, the thesis analyzed several of classical algorithms from the perspective of optimization theory, focused on the phase retrieval algorithm constraints, and summarized the number of non-zero pixels(NNP) constraints. On this basis, a hybrid iterative phase retrieval algorithm, which combines the advantages of each algorithm, was proposed. Since the algorithm recover support information by using the number of non-zero pixels constraint, it can greatly reduce the difficulty in obtaining the support in existing algorithms. Under the situations that the spectral amplitude of a sparse phase object is only known, the algorithm can recover the phase and amplitude of the original object. Extensive simulations showed the feasibility of the algorithm.(2) A ciphertext-only attack scheme for classic double random phase encoding encryption system was proposed. The scheme simply utilized the information of intercepted ciphertext, then the plaintext and deduced system key can be recovered effectively by using the iterative phase retrieval algorithms presented in this thesis. The scheme didn't need any prior information virtually, so it has been proven to be applicable. Compared with the existing ciphertext-only attack scheme, less iteration can produce higher correlation coefficient between recovered plaintext and original plaintext. Related simulation results demonstrate the validity and reliability of the scheme.