Home – Home – Jornals – Avtometriya 2026 number 4

Cross-view geolocalization is the process of establishing correspondence between images of ground objects obtained by surveillance equipment deployed on different platforms, provided that the images generated by the equipment on one platform are geodetically referenced. A cross-view geolocalization method is proposed based on the detection of landmarks e field of view of an unmanned aerial vehicle (UAV) video camera using the YOLOv12 neural network, trained on georeferenced satellite images containing these landmarks. The training set is formed by augmenting the images of the specified landmarks. The results of experiments obtained using a sample from the SUES-200 dataset containing 20 satellite images of various landmarks and 20×50 images of these landmarks obtained by a UAV onboard camera from an altitude of 300 m are presented.

This paper examines the design of algorithms that minimize the localization time of an unknown signal source that manifests itself by generating ultrashort (essentially instantaneous) pulses at random moments. The source being searched for is random in two respects: first, its position within the search interval is unknown, and second, the source generates pulses at random moments. The source is localized using a detector receiver capable of arbitrarily changing its viewing window at any moment in time. Optimal search algorithms are constructed for the case where the a priori probability density function of the source distribution is a two-level piecewise constant function. The optimal parameters of these algorithms are determined, including the expected execution time for each of them. It is proven that an optimal algorithm that takes into account a priori information about the intensity of the source being searched for is always more efficient than an algorithm for which this intensity is unknown.

Assessing the big data informativity is a topical area of research related to realizing the potential of big data technology for solving problems of automated monitoring, forecasting, decision-making, and machine learning in economic sectors. This article analyzes modern approaches and methods defining the concept of informativity and allowing for assessing the informativity in the context of the pragmatic and relativity of information, its value, and usefulness. The specifics of processing distributed and streaming big data and determining the minimum information space are considered. Assessing the big data informativity for problems of feature generation and selection, design, and data-driven management are discussed in detail.

A method for visualizing translucent objects with subsurface light scattering in real time is proposed. An analytical model is described that takes into account the directivity of the incident light, providing images that are more realistic compared to those generated using known analytical models. A sampling scheme based on the optical properties of the material is considered. A directional dipole using a rendering pipeline on a GPU is presented.

Statistical characteristics of the envelope ratios of two non-Gaussian signals and a mixture of two random processes with given distribution laws were determined. One-dimensional Density of Probability Distribution (DPD) for the ratio of envelopes of two independent non-Gaussian signals during incoherent processing, joint DPD of envelopes and their derivatives, as well as the average number of outliers were found. It is shown that one-dimensional DPD depends on the parameters characterizing the shape, scale and center of distribution of signal envelopes. Special cases of DPD calculations for Rayleigh and Rice distributions are given. Statistical characteristics of the polarization parameters of the signal, such as the DPD and the average number of emissions of the ellipticity coefficient, are obtained. Statistical characteristics are defined for the mixture ratio of two random processes with a normal distribution law and with a generalized non-Gaussian distribution. In particular, expressions were obtained for determining the DPD of a mixture of the two above processes, v-initial moments and an integral distribution function of the mixture of processes were found. The expressions DPD of the envelope of the useful signal in the presence of normal interference, DPD of the ratio of two statistically independent mixtures, DPD of the polarization phasor module, joint DPD of the mixture of two independent stationary processes and the average number of emissions are obtained.

Universal analytical expressions are proposed for calculating the quality factor (Q-factor) of optical amplitude-phase channel signals for data centerinterconnection (DCI) using Coherent Orthogonal Frequency Division Multiplexing (CO-OFDM). A single-span fiber-optic transmission system with an optical amplifier located at the receiving end and using polarization multiplexing for transmitting OFDM signals is considered. The resulting formulas take into account the simultaneous effects of nonlinear phase distortions occurring in the fiber and amplified spontaneous emission noise. These formulas enable the noise immunity of individual signal points in the M-QAM constellation to be estimated depending on the transmission system parameters, without resorting to resource-intensive simulation modeling. Calculations of the noise immunity of polarization-multiplexed 16-QAM channel signals showed the possibility of organizing a connection between Data Centers over a standard single-mode fiber over a distance of 80 km with a total data transfer rate of 32 Tbit/s.

The charge traps formation was studied in SiO₂ films implanted with In⁺ and As⁺ ions during subsequent annealing at temperatures T = 800-1100 °C. It was found, as the annealing temperature increased, the flat band voltage shifted non -monotonous and reached its maximum at the temperature 900 °C. At the same time, the hysteresis observed in the high-frequency C-V-characteristics was minimal. As the annealing temperature grew to 1000-1100 ^oC, the hysteresis window increased by an order of magnitude. From the obtained C-V-characteristics, the density of the fixed charge and the density of rechargeable charge carrier traps in the oxide was estimated. From the C-V and G-V-characteristics as a function of frequency, the surface state density was estimated, which was practically independent of the annealing temperature and was about 2×10¹¹ cm^-2eV^-1. The nature of the fixed charge in the dielectric was related to the oxygen vacancies In-As in the SiO₂ matrix. The increase in the hysteresis of the C-V-characteristics as a function of the annealing temperature was explained by a change in the energy spectrum of deep levels.

The potential of a method for determining protein concentration in biological fluids using Raman spectra of the dry residue after water evaporation is investigated. The method is used when the protein content in the liquid is low compared to non-protein components. To suppress contributions to the Raman spectrum from non-protein organic substances, protein precipitation with trichloroacetic acid (TCA) in the liquid before obtaining the dry residue is proposed. To obtain a quantitative value for the mass fraction of protein, it is proposed to add lysozyme, a protein with unique Raman spectral properties, to the liquid. It is shown that analysis of the Raman spectra of the dry residue with lysozyme precipitated with TCA allows determination of the mass fraction of protein in the range of 0.03 mg/ml and higher. The method is also tested using conditioned fluids after culturing corneal fibroblasts and myofibroblasts, demonstrating the feasibility of obtaining reliable protein concentration values.

Surface-enhanced infrared absorption (SEIRA) spectroscopy, which is based on the enhancement of the infrared response of an analyte in the vicinity of plasmonic nanoantennas, is a promising technique for highly sensitive molecular detection. The development of approaches to further improve the sensitivity of this technique remains an important challenge in modern sensor technology. In this work, arrays of linear gold nanoantennas fabricated on a silicon substrate were used to detect the characteristic vibrations of monolayer coatings. 1-Octadecanethiol (ODT) was employed as a model analyte, and a layer-by-layer coating procedure based on the Langmuir-Blodgett method was implemented. Two ODT monolayers were sequentially deposited onto the Au nanoantennas, allowing the absorption enhancement factor of the ODT vibrational band at approximately 1470 cm⁻¹, assigned to the bending vibrations of CH_₂ groups, to be estimated as 3×10⁵. The obtained results demonstrate the potential of plasmonic nanoantennas for the development of mid-infrared sensors.

A three-channel static polarimeter scheme designed for distributing spatial phase light modulators based on liquid crystals is presented. The interaction of light reflected from the modulator with the device's optical elements is calculated, and analytical expressions for the light intensity in each channel are obtained. A solution to the inverse problem of determining phase transitions on the matrix elements is presented. Experimental results demonstrating the transformation of a control signal into a phase transition are obtained.

Three metrological indicators are analyzed, the experimentally recorded values of which turned out to be inconsistent with the requirements of the three basic goniometry methods that formed the basis for the implementation of the selected measurement procedures.It is shown that the efficiency of the process of filtering the harmonics of the instrumental error in the reference sensor of the angular comparator, implementing one of the variants of the multi-combination method, turned out to be an order of magnitude lower than its theoretical value.It was shown that the results of calibration of the precision encoder RON-905, implemented using the mentioned comparator, were not in accordance with the requirements of the EDA method of the goniometry. It is also shown that the calibration results of one of the reading heads of the reference sensor of this comparator are distorted by the action of harmonics, the presence of which does not correspond to the original prescriptions of the self-calibration method.

The widespread use of carbon dioxide (CO₂) lasers in material processing, polymer welding, and lithography has led to a significant reduction in their cost, enabling their applications in new areas. In this work, we characterized the pulsed operation mode of a commercial Reci W2 sealed CO₂ laser tube that generates mid-infrared (mid-IR) radiation with an average power above 100 W. By combining pulse energy measurements with time-resolved FTIR spectroscopy, we examined the temporal profiles of the emitted radiation in the 10-1000 µs range and analyzed the spectral dynamics. Complementary measurements of current and voltage provided additional information on the discharge behavior and its temporal relationship with the optical output, offering a more complete picture of pulse formation. Two distinct modes of pulsed operation was demonstrated: 1) a transient high-peak-power regime achieved with relatively short 50-150 μs laser pulses, 2) a stable-generation regime observed for 150-1000 μs and longer laser pulses. The results show that inexpensive sealed CO₂ laser tubes can operate as convenient mid-IR pulsed sources with characteristics suitable for a range of spectroscopic measurements, materials research, and instrumentation development.

The work solves the problem of quantitative modeling of the semantic structure of text, which is formulated as the inverse problem of wave optics. Based on the established operational analogy between semantic dynamics and the propagation of coherent radiation in an inhomogeneous environment, geometrization of text in Minkowski space is proposed. Within this analogy, the concept of "semantic refractive index" is introduced, and the text is considered as a field of complex amplitude. The key result is to demonstrate the applicability of numerical methods of computer optics - the Finite-Difference Time-Domain (FDTD) method for direct modeling of the propagation of the semantic front and iterative phase recovery algorithms (Gershberg-Saxton) for reconstructing the latent thematic structure by the distribution of the "intensity" of embeddings. Experimental testing on a scientific text confirmed that the proposed approach makes it possible to visualize thematic clusters as a mode structure of a waveguide beam and quantify the coherence of discourse through a coefficient similar to the degree of spatial coherence. The work expands the scope of computer optics methods to the problems of analyzing complex semantic systems.