Optical Memory and Neural Networks最新文献

Based on the modeling of optical edge selection, studies were conducted to optimize the parameters of spatial filters. The parameters of the ring and vortex spatial filters were optimized for optical edge selection in different images using the Fourier-correlator. The result of the research is the development of an algorithm that helps to find the optimal parameters of ring and vortex spatial filters for a given image. The optimality criterion is determined using Canny’s digital algorithm, therefore, the developed algorithm is a hybrid algorithm. The developed algorithm for finding optimal filter parameters is based on minimizing the standard deviation from the samples formed by the Canny operator. The purpose of the developed algorithm consists of adjusting the Spatial light modulator to the input image in a real experiment. The fact that the vortex filter with optimal parameters is slightly inferior to the Canny’s operator was established. It has been shown that the vortex filter makes it possible to selectively highlight object details.

A matrix formalism, denoted as ABCD, has been developed and experimentally implemented to describe structured transformations and the orbital angular momentum of structured Laguerre–Gaussian beams in an astigmatic optical system containing a single cylindrical lens. It is demonstrated that the matrix formalism not only aligns well with the method of integral astigmatic transformations but also significantly broadens the scope of its application.

The possibility of recognizing wave aberrations using a convolutional neural network with the Xception architecture is investigated based on intensity patterns at the output of a Fourier correlator with a multichannel spatial filter matched with Zernike basis functions. A dataset was calculated for training a neural network. In this dataset the intensity distribution at the correlator output was modeled for each of the first eight aberration types and their superpositions. Based on network training in 80 epochs, it was found that for the validation sample, the mean absolute error in recognizing aberrations does not exceed 0.003.

Multiplexing a signal using the largest number of freedom degrees is an important task in the field of optical information transmission. Variations in the beams amplitude-phase structure can provide a wide range of possibilities for encoding and multiplexing optical signals and do not require complex optical systems or expensive devices. The paper analyzes the results of focusing laser radiation using phase vortex shifted lens quantized at a given level. It is shown that a shifted quantized lens makes it possible to obtain a variable number of local intensity maxima out of the optical axis. The use of a shifted quantized vortex lens made it possible to register a useful signal and detect both individual vortex beams and their superpositions out of the optical axis. We can also note that studies of the quantization parameter influence on the propagation of beam intensity have shown the possibility of eliminating fixed diffraction orders. Individual quantization levels produce diffraction orders that account for equal shares of energy (weights), and are thus candidates for more complex detection tasks that also include weight recognition.

The present paper is concerned with the analytical and numerical investigation of the dynamics of a broad-area vertical-cavity surface-emitting laser (VCSEL). It is shown that the adiabatic polarization elimination does not affect the modulation instability developing in the system. The efficiency of the method of external injection at different shapes of the pump current profile is verified. It is shown that for effective stabilization by optical injection beam with minimum amplitude it is necessary to use pump current profiles with smoother edges.

The description of the role of absorption in interference resonators is important for characterization of metal-insulator-metal structures. To study the resonance characteristics of the interference structures the conditions of excitation of their eigenmodes were analyzed. The mode excitation conditions determined by the dispersion relations were obtained for the waveguide, Fabry-Pérot, symmetric and antisymmetric plasmonic normal modes. The approximations of the dispersion relations in the form of analytical functions were implemented to describe the behavior of the resonance positions in far-field spectra depending on the parameters of the layers.

We investigate the spin angular momentum of a superposition of two vector light beams with rotational symmetry. One beam is cylindrically polarized and another is linearly polarized. Radial form of these beams can be arbitrary (Laguerre-Gaussian, Bessel-Gaussian, or some other). For such a superposition, an analytical expression is derived for the spin angular momentum and two its properties are obtained. At first, we found that altering the coefficients does not affect the form of the spin angular momentum distribution, while the intensity distribution is changed. At second, we found that maximal spin angular momentum is generated if both beams are of equal power.

Various types of diffraction and plasmonic elements are discussed that provide modern development of sensors in different applications such as analysis of the mode composition of radiation and wavefront aberrations based on multichannel filters, detection of the polarization state using diffraction microstructures, as well as biochemical analysis using metal-insulator-metal waveguides. The considered types of sensors are promising in the field of optical communications, quantum information, materials testing, environmental monitoring.

Topological charge is an important characteristic of a laser beam and is closely related to orbital angular momentum. Although in most cases the charge is conserved when a beam propagates in space, the opposite examples are known. We formulate and prove a general theorem for conservation of topological charge, which involves not only charges in the plane perpendicular to the propagation axis, but also in planes parallel to the propagation axis. Charges of the second type can be identified to some approximation with spatiotemporal charge, which have been widely studied in recent years.

The study presented in this paper is devoted to the development of technology for detecting osteoporotic changes in vertebral bodies according to computed tomography data based on the analysis of groups of textural features. This technology will be of interest to radiologists when assessing the structure of vertebral bodies in patients with osteoporosis, and especially in patients with malignant neoplasms, in which the disease proceeds with a decrease in bone mineral density. Automation of the analysis process will significantly reduce the time spent by a radiologist on the evaluation of vertebral bodies, including when evaluating the effectiveness of osteoporosis treatment. During the development process, the most informative group of signs was identified, which allows achieving high accuracy in detecting signs of osteoporotic changes in the vertebral bodies in 94.5%.