Workshop on Information Optics 

Abstract: A novel technique for multiplexing complex images is proposed in which each image may be demultiplexed only if a set of random encryption keys is known. The technique utilizes the ability of the double random phase encoding method to spread a signals’ energy in both the space and the spatial frequency domains in a controlled manner. To multiplex, images are independently encrypted with different phase keys and then superimposed by recording sequentially on the same material. Each image is extracted by using the particular key associated with it. During decryption the energy from the other images is further spread, making it possible to minimize its effects by using suitable filters. Wigner analysis is applied to the technique, and numerical results are presented. © 2007 Optical Society of America OCIS codes: 070.6020, 090.4220, 070.2580.

Abstract: The method of sub‐Nyquist sampling is considered that is used to decrease sampling speed in a few times for narrow‐band OCT signals. To provide high noise‐immunity and stability when processing OCT signals with randomly variable parameters in real time, the Kalman filtering method has been applied for evaluating sub‐Nyquist sampled OCT signals.

Abstract: Augmented reality aims to mix real-world visual content with virtual objects. Achieving realistic results involves solving challenging computer vision tasks, such as tracking real 3D objects and estimating the illumination conditions of a scene. In this short paper, we present how these two challenging tasks can be solved robustly and accurately with deep learning. In both cases, deep convolutional neural networks are trained on large amounts of data, and achieve state-of-the-art results.

Abstract: We overview a compact and field-portable digital holographic microscopy system based on shearing geometry and integrated with a head mounted augmented reality device for cell identification and visualization. Customized smart glasses containing an external camera connect directly to the 3D-printed system to record holograms of biological specimens. Following hologram acquisition, regions of interest containing biological cells are segmented and digitally reconstructed to generate a three-dimensional (3D) pseudocolor optical path length profile. From the optical path length profiles, morphological features are extracted and inputted into several classification models for comparison including random forest classifier, support vector machines, and k-nearest neighbor, each yielding a high classification accuracy. After successful classification of the target cell, the classification result along with a pseudocolor 3D rendering of the cell’s optical path length profile, and its extracted feature values are displayed to the augmented reality device for the user. The system was tested on both living and non-living samples, including feature extraction from video data of live paramecium. The overviewed system may allow for quickly and conveniently visualizing cells through an augmented reality device and extracting relevant information with potential applications of rapid diagnosis by healthcare professionals working in remote areas. We acknowledge support from the National Science Foundation (Directorate for Engineering (ENG) (NSF ECCS 1545687, NSF/IIS-1422179)).