A lensless optical security system based on double random-phase encoding in the Fresnel domain is proposed. This technique can encrypt a primary image to random noise by use of two statistically independent random-phase masks in the input and transform planes, respectively. In this system the positions of the significant planes and the operation wavelength, as well as the phase codes, are used as keys to encrypt and recover the primary image. Therefore higher security is achieved. The sensitivity of the decrypted image to shifting along the propagation direction and to the wavelength are also investigated.

Since their inception in the 1930-1960s, the research disciplines of computational imaging and machine learning have followed parallel tracks and, during the last two decades, experienced explosive growth drawing on similar progress in mathematical optimization and computing hardware. While these developments have always been to the benefit of image interpretation and machine vision, only recently has it become evident that machine learning architectures, and deep neural networks in particular, can be effective for computational image formation, aside from interpretation. The deep learning approach has proven to be especially attractive when the measurement is noisy and the measurement operator ill posed or uncertain. Examples reviewed here are: super-resolution; lensless retrieval of phase and complex amplitude from intensity; photon-limited scenes, including ghost imaging; and imaging through scatter. In this paper, we cast these works in a common framework. We relate the deep-learning-inspired solutions to the original computational imaging formulation and use the relationship to derive design insights, principles, and caveats of more general applicability. We also explore how the machine learning process is aided by the physics of imaging when ill posedness and uncertainties become particularly severe. It is hoped that the present unifying exposition will stimulate further progress in this promising field of research.

We introduce the technique of wavelength multiplexing into a double random-phase encoding system to achieve multiple-image encryption. Each primary image is first encrypted by the double phase encoding method and then superposed to yield the final enciphered image. We analyze the minimum separation between two adjacent multiplexing wavelengths through cross talk and the multiplexing capacity through the correlation coefficient. Computer simulations are performed to demonstrate the concept. This technique can be used for hiding multiple images as well.

In this manuscript, we propose a novel framework of computational ghost imaging, i.e., ghost imaging using deep learning (GIDL). With a set of images reconstructed using traditional GI and the corresponding ground-truth counterparts, a deep neural network was trained so that it can learn the sensing model and increase the quality image reconstruction. Moreover, detailed comparisons between the image reconstructed using deep learning and compressive sensing shows that the proposed GIDL has a much better performance in extremely low sampling rate. Numerical simulations and optical experiments were carried out for the demonstration of the proposed GIDL.