| Optical information security is becoming a state-of-the-art cross-discipline topic nowadays. An optical system is characterized by the powerful capability of information processing parallely across three dimensions and advantageous at providing multiple parameters regulating a propagation, such as a wavelength, diffraction distance, amplitude and phase. These advantages guarantee the information encryption and hiding by optical processing to perform at multiple dimensions with a mass-storage, and parallel information processing, apparently over conventional information security techniques. Recently, a great number of optical information security techniques for multiple images mushroom, which results in the fact that these techniques can be translated into applications of content distribution, multiple-user authorization and confidential information transferring and storing.The thesis summarizes the progress of optical information security and introduces advances of image encryption by optical system and techniques. Based on these widely used optical encoding systems and hiding technologies, we proposed to encrypt and hide multiple images compressively and meanwhile resolve key related problems systematically and completely. In the thesis, we started with the fundamental analysis of optical information processing to verify the feasibility and performance. Then the computational optics was involved to introduce the techniques originated from optical information processing into the application, such as specific domain hidden in optics. The physical optical system was used to demonstrate the algorithms and performance in terms of parallel processing in optics. Major contributions covers four aspects as follows,1) Multiple images are scanned and recorded simultaneously based on digital holography simulating optical scanning holography. The encryption is further improved by multiplexing the diffraction distance and applying masks to shuffle the distribution of complex amplitude of digital holograms. The decryption is modeled by an inverse problem and the solution is optimized by the Tikhonov regularization. The results from numerical experiments demonstrates the feasibility, robustness and security of the algorithm and the reconstruction of images is advantageous over existing algorithms.2) Taking advantage of compressive hologram, it is roposed to compress encrypted holograms by non-uniform down sampling. In the algorithm, we deduce the optical transfer function of the optical scanning holography and reveal the function analogous to certain kind of Fourier transform. Furthermore, owing to the sparsity of natural images under wavelet decomposition and the incoherence between wavelet sparsity operator and Fourier operator, a non-uniform down sampling of Fourier holograms is guaranteed to reconstruct original images based on the compressive sensing theory. Finally, the reconstruction is implemented by solving an inverse problem with a sparse constraint.3) Total variation is involved to reconstructing imaged with fine edges and wavelet processing is employed as a pre-processing step to record information of multiple images. The above algorithm lays a foundation for an encryption strategy by a compressive hologram. Here the solving operation is optimized with a total variation to achieve images with edges in replace of a12norm. In order to reduce the cross-talk effect during scanning multiple images, the wavelet decomposition is utilized as a pre-processing step to avoid overlapping of multiple image information, which ensures the information of images recorded completely. Not only is the algorithm demonstrated by a simulation, but also it is implemented in a physical optical scanning holography system. These experiments verifies the algorithm, investigate the compressive capability and reconstruction quality.4) The phase retrieval algorithm realizes the multiple-image hiding with less lenses. Phase filters are involved in the algorithm to hide several image by frequency multiplexing. It outputs an image containing all frequency information of images. The phase retrieval method is introduced to find a specific phase mask to modulate the image similar to the carrier image and we take the correlation coefficient as the threshold. Finally, the carrier image is displaced into the right position and original secret images are extracted with the corresponding mask and phase filters. Numerical experiment demonstrates the feasibility of the algorithm and analyze the recognition of hiding information and the convergence of phase retrieval algorithm. |
PDF Full Text Request