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.

Quality assessment and artifact detection in functional magnetic resonance imaging (fMRI) data is essential for clinical applications and brain research. Subject head motion remains the main source of artifacts -even the tiniest head movement can perturb the structural and functional data derived from the fMRI. In this paper, we propose an end-to-end neural network technology for detecting step anomalies with training on partially synthetic data with adaptation to a specific small set of real data. A procedure for generating a synthetic dataset for training and a module for automated labeling of real data is developed. A recurrent neural network model for detecting step anomalies is proposed. A method for the model adaptation to a small set of real data based on onestep metalearning is developed. An experimental verification of the accuracy is carried out in the problem of detecting step anomalies using a sliding window of 10, 15, and 24 pixels. The experiments have shown the proposed technology to provide the detection of stepwise anomalies with an accuracy of 0.9546.

Optical signal processing components, such as optical modulators, sensors, all-optical integrators, and differentiators implemented on a chip, are important for developing computer technology. Analog signal-processing applications include all-optical solution of differential equations of various orders. Integrators and differentiators based on Bragg gratings are a few millimeters in size. Integrators and differentiators based on ring resonators are more compact. The model of a two-component nanocavity is also described. In this model, the minimum details of the structure are found only in the periodic component of the resonator. The optical modulator is one the most important optical signal processing components. The optical resonant sensors demonstrate good compatibility with the requirements of the microfluidics. The resonance cavity characteristics were computed using the parallel three-dimensional (3D) finite-difference time-domain method. The resonance cavity regions that contain the nonlinear material are characterized by a smaller refractive index than that of the waveguide.

Modern methods of computational photography make it possible to bring the quality of images obtained by mobile cameras closer to the quality of professional cameras. One of the most important tasks is that of ensuring the consistency of colors from different cameras. In this paper, we propose a simple and efficient way to bring the colors of one camera to another, based on the approximation of the required transformation by a tone correction spline and a color transformation matrix. An experimental study was carried out in a rather complicated case, in which it was required to match colors of the images obtained from two fundamentally different sensors, as well as using diffractive optics. The results of the experiments showed that the proposed method allows one to obtain a higher accuracy of color matching between cameras than existing analogues.

Recently, data mining and neural networks are increasingly used for wavefront recognition from interferograms. In this case, there is considerable freedom in choosing the structure of the reference beam. In this work, a comparative study of the effectiveness of using neural networks for solving the problem of recognizing wavefront aberrations based on linear (flat reference beam) and conical (conical reference wavefront) interferograms is carried out. The effectiveness of recognition of types and levels of aberrations by conical interferograms based on the use of neural networks is shown: the average absolute error is reduced by 3 times, compared with linear interferograms. This effect is related to the rotational invariance of the introduced aberrations.

Synthetic Aperture Radar (SAR) interferometry is an active remote sensing technology that uses microwaves to characterize the earth's surface. SAR interferometry allows to measure the 3D profile of the earth's surface, recover surface topography, and determine topographic displacements over time. The microwave SAR signal is usually highly distorted. Distortions can be caused by, for example, atmospheric disturbances and various characteristics of earth's surface scatterers reflectance. Compensation for these distortions is performed by filtering the phase and evaluating the degree of coherence of the original images. This is an important step to improve the accuracy of the subsequent pphase-unwrapping operation. In this paper, we investigate the use of U-net neural networks for preprocessing the SAR interferogram at various parameters of the distortion of the SAR signal. Two neural networks filter the SAR interferogram and determine the degree of coherence, respectively.

Recently, intelligent data analysis and neural networks are increasingly used to detect the wavefront by interferograms and digital holograms. In this case, there is significant freedom to choose the structure of the reference beam. In this paper, a comparative study of the effectiveness of the neural network was performed to solve the problem of the wavefront aberration recognition based on off-axis and inline schemes of digital holography with a plane and conical reference wavefront, respectively. The feature of the inline digital holograms with the conical wavefront compared to the off-axis is the invariance of their structure to the rotation of the wavefront at some angle. In addition, a numerical analysis of the sensitivity of the types of digital holograms under consideration to changes in the wavefront for a different level of aberration showed relatively better characteristics for the average level of aberration, when it is difficult to apply detecting methods based on the Shack-Hartmann sensors or matched filtering. These features of inline digital holograms with a conical reference wavefront made it possible to increase the recognition efficiency for types and levels of aberrations using neural networks. As a result, the average absolute recognition error for model interferogram decreases more than three times. The results for the experimental conical and linear interferograms turned out to be quite close because of the sensitivity of the conical wavefront to the alignment of the optical system. Moreover, the neural network trained on a more diverse experimental data set, which contains natural distortions of image registration, gives an increase in the average recognition accuracy for linear-type interferograms. Thus, in the future, it is reasonable to consider the combined use of various types of interferograms.

The paper proposes a video surveillance scheme for compact placement of a system for railway rolling stock accounting. This design is based on the use of a tilted diffractive optical element and a tilted lens. Such an optical design makes it possible to significantly increase the depth of focus of the imaging system. This work considers the influence of the tilt of a diffractive lens on the shape and size of the focused area. Analytical relations describing the geometry of the focused region for various spectral channels are given. The possibility of increasing by several times the size of the zone of accurate image classification using a neural network has been demonstrated. The proposed approach has been tested on real-world dataset of images of house number plates.

Recognition of the types of aberrations corresponding to individual Zernike functions were carried out from the pattern of the intensity of the point spread function (PSF) outside the focal plane using convolutional neural networks. The PSF intensity patterns outside the focal plane are more informative in comparison with the focal plane even for small values/magnitudes of aberrations. The mean prediction errors of the neural network for each type of aberration were obtained for a set of 8 Zernike functions from a dataset of 2 thousand pictures of out-of-focal PSFs. As a result of training, for the considered types of aberrations, the obtained averaged absolute errors do not exceed 0.0053, which corresponds to an almost threefold decrease in the error in comparison with the same result for focal PSFs.

The impact a diffractive lens tilt on the parameters of the focal plane for the image classification problem is studied. Analytical expressions for the geometry of the focal planes for different color channels are derived. Image classification tests are performed on a completely simulated framework. Convolution of source image and the point spread function (PSF) is used to model the blurred image at the output of the imaging system. In this simulation the PSF is depends on the depth map of the source image. The work demonstrates the possibility of expanding by several times the size of the region of accurate image classification employing a convolutional neural network. The paper results can be utilized in robots machine vision modules and drones.

This work proposes a method for increasing the accuracy of wavefront aberration recognition. The method involves the sequential use of two convolutional neural networks. The first neural network classifies the input point spread function (PSF) images according to the type of aberration. The second neural network performs the functions of a regressor, i.e. evaluates the values of the Zernike coefficients. It is shown in the work that this approach allows to increase the accuracy of calculating the values of the Zernike coefficients by approximately one and a half times in comparison with the previous approach.

Using an example of a real-world data set, it is shown that the accuracy of the image detector based on a YOLOv3 neural network does not deteriorate when using only one nonblurred color channel. The binary diffractive optical element was calculated, which allows increasing the imaging system depth of field by several times. This is achieved by using different color channels for various defocus values. A comparison of the MTF curves of the original and apodized imaging systems for a given minimum acceptable value of image contrast is presented. This approach allows us to create novel remote sensing imaging systems with an increased depth of field.

We performed a detailed comparative study of the parametric high degree (cubic, fourth, and fifth) power phase apodization on compensation defocusing and chromatic aberration in the imaging system. The research results showed that increasing the power degree of the apodization function provided better independence (invariance) of the point spread function (PSF) from defocusing while reducing the depth of field (DOF). This reduction could be compensated by increasing the parameter α; however, this led to an increase in the size of the light spot. A nonlinear relationship between the increase in the DOF and spot size was shown (due to a small increase in the size of the light spot, the DOF can be significantly increased). Thus, the search for the best solution was based on a compromise of restrictions on the circle of confusion (CoC) and DOF. The modeling of color image formation under defocusing conditions for the considered apodization functions was performed. The subsequent deconvolution of the resulting color image was demonstrated.

The binary optical element for phase apodization of the pupil function of the optical system is calculated in the paper. This optical element provides an increase in the depth of field of the optical system through at least one color channel for a certain value on the optical axis. The optimization of the binary optical element was performed by simulating annealing. It is shown that the calculated optical element provides an increase in the depth of field of the optical system by about two times compared with an optical element based on a binary axicon. The calculated optical element can be used in machine vision problems for image classification.

Linear accelerators are complex machines that can face significant periods of downtime due to anomalies and the subsequent failure of one or more components. The need for reliable linear accelerator operations (LINAC) is critical to the spread of this method in the medical environment. At CERN, where LINACs are used for fundamental research, similar problems are encountered, such as the appearance of jitter in plasma sources (2 MHz RF generators), which can have a significant effect on subsequent beam quality in the accelerator. The SmartLINAC project was created to increase LINACs' reliability by early detection and prediction of anomalies in its operations, down to the component level. This article shows how anomalies were first discovered and goes deep into understanding the nature of the data. The research adds new elements to anomaly detection approaches used to record jitter periods on 2MHz RF generators.

Using an example of a real-world data set, it is shown that the accuracy of the image classifier based on a convolutional neural network does not deteriorate when using only one color channel. The binary diffractive optical element was calculated, which allows to increase the imaging system depth of field by several times. This is achieved by using the different color channels for various defocus values. A comparison of the MTF curves of the original and apodized imaging systems for a given minimum acceptable value of image contrast is presented.

The paper proposes using a two-zone different level Fresnel lens to increase the depth of field. On the one hand, such a diffractive optical element can reduce the weight of the device compared to, for example, cubic phase and binary axicon apodization. On the other hand, such an element has a simpler structure compared to a harmonic lens or free-form DOE. A neural network is used to restore the image. Optimization of the surface relief of the proposed two-zone lens is performed.

INTRODUCTION:In 2014 and 2015 Professor of neurology Andrey Bryukhovetskiy published a novel theory of the information-commutation organization of the human brain in Russia, China and the USA. The theory posits the hypothesis that the higher nervous activity (cognitive, intellectual, mnestic) of the humans and their mind are material and have microwave electromagnetic nature. The theory perceives the human mind as a result of dynamic extracortical information-commutation relations of the super-positions of the electromagnetic waves of ultra high frequency emitted by different areas of the human brain in the inter-membrane cerebrospinal fluid space of the human head at a certain period of time. The inter-membrane cerebrospinal fluid space of the human head (the space between the dura, arachnoid and pia mater filled with the cerebrospinal fluid) of about 10mm size, has all morphological attributes to realize the holography. It is a universal natural bioprocessor for processing, analysis and synthesis of the input data and their record or reproduction on the pia as on the biological holographic membrane. The theory suggested that the processes of the mind can be recorded and digitalized with the last generation contemporary microwave receptors of the UHF band. GOAL:The goal is to experimentally test the theory of the information-commutation organization of the human brain, particularly, the postulate that the human mind has material, and, namely, electromagnetic nature represented by the microwave bioelectric activity; it must be detected, recorded and statistically processed, i.e. its existence must be confirmed. METHODS:On their own initiative, the team of mathematicians, radioengineers and neurologists performed the non-invasive research of the electromagnetic radiation of human brain in the broad frequency range varying from 850MHz to 26.5GHz with the last generation specialized measuring equipment with high sensitivity and recording speed, specialized measuring antennas and low noise amplifying equipment in the anechoic chamber of the 1st class of protection according to the Russian system of certification GOST R 50414-92. RESULTS:The previously unknown microwave electromagnetic radiation of the EHF/UHF range (from 1.5GHz to 4.5GHz) with signal strength of -130dBm .. -100dBm (1e-15 .. 1e-13 W) are discovered. The detected electromagnetic waves have zonal variations in the different areas of the human head and are absent in other areas of the human body. The method of recording of the microwave electromagnetic activity of the human brain is patented in the Russian Federation. The microwave electromagnetic activity of the brain is billion-fold different from the bioelectric activity recorded by the encephalography. CONCLUSION:Discovery of the phenomenon of the microwave radiation of the human brain provides evidence to the idea that thinking and mind are material. This phenomenon has the potential to become a new informational channel of the diagnostics of the functional and pathological state of the higher nervous activity of the human brain. It can provide the basis for the development of the equipment for real-time analysis of the microwave bioelectric activity of the brain in norm and pathology, for objective early diagnostics of the functional and emotional conditions as well as of the psychiatric disorders at the preclinical stage, for the biocontrol of the human brain and the artificial simulators of the human brain. It also can provide the foundation for new systems of the artificial intellect, brain-computer interface and systems of the closed-loop biomanagement of the damaged brain.

This work considers the influence of the tilt of a diffractive lens on the shape and size of the focused area. Analytical relations describing the geometry of the focused region for various spectral channels are given. The possibility of increasing by several times the size of the zone of accurate image classification using a neural network has been demonstrated. The results of the work can be used in machine vision units for robots and unmanned aerial vehicles.

In this work, we carried out training and recognition of the types of aberrations corresponding to single Zernike functions, based on the intensity pattern of the point spread function (PSF) using convolutional neural networks. PSF intensity patterns in the focal plane were modeled using a fast Fourier transform algorithm. When training a neural network, the learning coefficient and the number of epochs for a dataset of a given size were selected empirically. The average prediction errors of the neural network for each type of aberration were obtained for a set of 15 Zernike functions from a data set of 15 thousand PSF pictures. As a result of training, for most types of aberrations, averaged absolute errors were obtained in the range of 0.012 – 0.015. However, determining the aberration coefficient (magnitude) requires additional research and data, for example, calculating the PSF in the extrafocal plane.