Research On Optical Imaging Techniques Based On CGH And Phase Coding

With the advancement of computers and holography,computer-generated holography(CGH)and phase encoding techniques have been widely used in the field of optical imaging.Especially in the field of optical information security,optical imaging technology based on CGH and phase coding has become an important research direction for information security processing.In recent years,researchers have proposed a variety of different imaging techniques and methods that can be used for information security encryption.At the same time,deep learning has emerged has achieved great success in the fields of image recognition and speech recognition.How to apply deep learning to optical imaging and optical encryption,and make it a powerful tool for solving difficult problems in those fields has become one of the research priorities in information security.In this thesis,we first present a method for generating authenticable phase-only hologram based on integrated two-stage phase optimization,and then describe an optical image encryption system in detail based on deep learning and holographic speckle imaging.Specifically,the main work presented in the thesis can be summarized as follows.1.We propose a novel method for generating an authenticable phase-only hologram(APOH).The integrated two-stage phase optimization algorithm can be used for holographic phase optimization and steganographic encryption.The APOH is calculated by the proposed integrated two-stage phase optimization under phase constraints are involved.The method has the following characteristics.First,the method provides a high security level.The original image,i.e.,hidden image,cannot be obtained from the CGH itself or its diffraction patterns,and the joint use of authentication key(A-key)and decryption key(D-key)increases the security level.Second,the quality of reconstructed images is improved,and the speckle noise and DC noise in the diffraction image are reduced by phase optimization.Third,high decryption efficiency can be achieved by the verification of CGH with the A-key,which provides a way to identify the CGHs in a secure manner.2.We propose an optical image encryption system based on deep learning and holographic speckle imaging.A steganographic scheme is used for encrypting a secret image into a synthetic phase CGH by using a hybrid non-iterative procedure.A modified Dense Net is applied to perform the decryption of secret images.When the synthetic CGH is illuminated by the same laser beam without decryption,the cover image is displayed so as to disorient the attacker,which also provides a simple way to match a synthetic CGH and decryption key pair.The speckle pattern diffracted by the CGH,which is decrypted from the synthetic CGH,is the only input to a Dense Net model,which is pre-trained to estimate the relationship between the secret images and noise-like diffraction patterns that were recorded optically.Experiment results demonstrate that the primary images can be successfully retrieved due to the nonlinear characteristics of deep neural networks,although the optically reconstructed images with a decryption key visually look like white noise for security reasons.