TELEVISION AND IMAGE PROCESSING
Introduction. Computer vision systems are finding widespread application in various life domains. Monocularcamera based systems can be used to solve a wide range of problems. The availability of digital cameras and large sets of annotated data, as well as the power of modern computing technologies, render monocular image analysis a dynamically developing direction in the field of machine vision. In order for any computer vision system to describe objects and predict their actions in the physical space of a scene, the image under analysis should be interpreted from the standpoint of the basic 3D scene. This can be achieved by analysing a rigid object as a set of mutually arranged parts, which represents a powerful framework for reasoning about physical interaction.
Objective. Development of an automatic method for detecting interest points of an object in an image.
Materials and methods. An automatic method for identifying interest points of vehicles, such as license plates, in an image is proposed. This method allows localization of interest points by analysing the inner layers of convolutional neural networks trained for the classification of images and detection of objects in an image. The proposed method allows identification of interest points without incurring additional costs of data annotation and training.
Results. The conducted experiments confirmed the correctness of the proposed method in identifying interest points. Thus, the accuracy of identifying a point on a license plate achieved 97%.
Conclusion. A new method for detecting interest points of an object by analysing the inner layers of convolutional neural networks is proposed. This method provides an accuracy similar to or exceeding that of other modern methods.
RADAR AND NAVIGATION
Introduction. Currently, there is a significant increase in the number of relay satellites in geostationary orbit. However, frequent incidents of illegal use of the satellites frequency resource, as well as unintentional and deliberate interference with other users are fixed. In this regard, it becomes necessary to evaluate accuracy and applicability of various methods for determining the location of sources of illegal and interfering radio emission with different signal parameters and with different levels of uncertainty for relay satellite coordinates and velocities.
Aim. To study and to evaluate the accuracy of methods of geolocation of radio emission sources operating through geostationary relay satellites, with different signal parameters and with different levels of uncertainty for relay satellite coordinates and velocities.
Materials and methods. Imitation modeling and the theory of digital signal processing were used.
Results. Factors influencing the accuracy of the estimation of TDOA and FDOA parameters when determining the position of radio emission sources, which operate via relay satellites located in geostationary orbit, were considered. As a result of simulation, the estimate of the accuracy of the considered geolocation methods was obtained. It depends on the bandwidth of radio emission source signal, on the recording duration and on the level of a priori uncertainty relatively the relay satellites coordinates and velocities. Recommendations for the application of the considered methods in various conditions were formulated.
Conclusions. Conclusions and recommendations formulated as a result of the study, will allow one to choose the most appropriate geolocation method to improve the accuracy of radio emission sources locating depending on conditions and signal parameters.
Introduction. The point and interval direction finding of radio sources is used for broadband radio monitoring in the frequency domain. The initial data for broadband radio monitoring are spectral samples obtained from an M-element antenna array by multichannel reception. Point direction finding is based on a grouping of point estimates of azimuth and elevation angle formed for each frequency sample, in which signal components are detected. A single estimate of azimuth and elevation angle is made based on the grouped point estimates in the range of neighbouring frequency samples. Interval direction finding is based on the azimuth and elevation estimates formed entirely from the interval of adjacent frequency samples, in which the signal components are found, and the subsequent refinement of frequency sample interval boundaries for each radio source in multisignal mode by spatial selection methods. Point direction finding is mainly implemented in single-signal mode in modern operating broadband radio monitoring complexes, while the multi-signal mode based on MUSIC or ESPRIT is implemented in the time domain in a narrow frequency band.
Aim. Development and investigation of methods for point and interval direction finding in multi-signal mode, as well as development of recommendations for their practical application in multi-signal and single-signal modes.
Methods. Multi-signal mode for point and interval direction finding was implemented using MUSIC and ESPRIT. An experimental study of the developed direction finding methods in single-signal and multi-signal (on ESPRIT) modes with overlapping signal spectra was carried out by processing the recorded real signals. The records were made using a seven-channel coherent synchronous receiver connected to a seven-element 60° angle antenna array.
Results. The research results are presented by frequency-azimuth panoramas and estimates of the amplitude spectra of separated signals and direction finding accuracy indicators.
Conclusion. It was experimentally demonstrated that point direction finding should be used in single-signal mode provided the absence of information on the number of signals in the observed data. Interval direction finding is recom-mended in multi-signal mode for improving the accuracy and real-time feasibility of the process.
Introduction. The possibility of application of modified parametric methods of spatial signal processing in a sparse antenna array (SEAA) of the receiving position of transportable over-the-horizon decameter range radar (DRR) intended for all-weather remote monitoring of the shelf zone is considered in this paper. With an operational deployment of DRR on unprepared coast, problems of the equidistant location of antenna elements (AEs) often arise. In the case of nonequidistant AEs location and matched spatial processing, antenna pattern has interference sidelobes, which level can significantly exceed the allowable or calculated one for an equidistant AA. A well-known alternative to matched processing are parametric methods of spectral analysis based on the using of models with a finite number of parameters, but their direct application requires an equidistant sampling of the spatial signal.
Aim. The aim of the research is to develop and analyze the method of parametric processing of spatial signals of the SEAA which AEs are located on the line with a random step in the range from λ/2 to several λ, where λ is the DRR wavelength.
Materials and methods. To construct the detection characteristics (DC) computer modeling in the MatLab environment, the reliability of which was confirmed by the construction of known and theoretically calculated DC, was used.
Results. The developed method includes a procedure of restoring (synthesizing) of artificial signal of equidistant AA with subsequent application of Burg parametric algorithm to obtain an estimate of the angular spatial frequency spectrum. To prove the applicability of the parametric method of SEAA signals processing in the case of location signals detecting, DC were obtained and compared with optimal ones.
Conclusions. The obtained results have proved the suboptimality of the parametric method of SEAA signal processing at the random AEs spacing step lying in the range from λ/2 to 3λ, what makes it possible to recommend it for using in transportable DRRs.
Introduction. Wide area multilateration (WAM) systems are the main competitors of secondary surveillance radar (SSR) systems used in air traffic control (ATC). The general principle of WAM operation is based on the assessment of pseudoranges between a signal source (an aircraft airborne transponder) and the ground receivers with precisely known geographical coordinates deployed over the ATC area. The aircraft position is estimated by measuring pseudoranges. A significant factor affecting the accuracy of aircraft positioning is tropospheric refraction, a phenomenon caused by the inhomogeneity of the earth's atmosphere and manifested in a deviation in the direction of the rays along which the signal of an aircraft transponder propagates. Refraction increases the lengths of ray paths, thus increasing the corresponding pseudoranges. As a result, the estimate of the aircraft position receives an additional bias. Altitude estimates produce unreasonably large errors.
Aim. To develop a mathematical model for the signals received by a WAM system, which accounts for tropospheric wave propagation, as well as to derive an algorithm for aircraft positioning with compensated tropospheric errors.
Materials and methods. Equations for the pseudorange estimation errors caused by wave propagation in a spherically stratified atmosphere were derived using the method of geometrical optics.
Results. This paper proposed a mathematical model for pseudorange estimates in WAM systems, which accounts for the bias associated with the phenomenon of tropospheric refraction. An analysis of the proposed model showed that pseudorange errors depend linearly on the distance between the aircraft transponder and the receiver. This conclusion allowed an algorithm for aircraft positioning with compensated tropospheric errors to be developed. The proposed algorithm yields an unbiased estimate of the aircraft position. The standard deviation of altitude estimates increases by 60%, although remaining within the limits permissible for WAM systems.
Conclusions. The developed mathematical model of WAM signals, which considers tropospheric propagation errors in pseudorange estimation, as well as the algorithm for aircraft positioning with compensated tropospheric errors, can be used in the development of spatially distributed navigation systems.
MEASURING SYSTEMS AND INSTRUMENTS BASED ON ACOUSTIC, OPTICAL AND RADIO WAVES
Introduction. At the present, sensors based on surface acoustic waves (SAW) is a rapidly developing direction and a promising replacement for classical sensors, especially in those areas where long-term performance of latter is questionable. The principle of operation of SAW sensors is based on acoustic vibrations, therefore, the choice of piezoelectric material of а console, considering external influences on a future device and its operating conditions, is the most important task. Currently, many monocrystalline structures and their sections have been synthesized and created for the devices using SAW. The main materials used for the manufacture of substrates are crystals of quartz (SiO2), lithium niobate (LiNbO3), lithium tantalate (LiTaO3) and film aluminum nitride (AlN). Also, new crystal structures: langasite (La3Ga5SiO14), langatate (La3Ga5.5Ta0.5O14), langanite and others were produced. The problem of using such materials for the manufacture of consoles is the lack of systematized data on important characteristics for the propagation of surfactants, for example, the elasticity tensor of the 4th rank. One of the key problems for the further development of SAW-based sensors is the one-way fastening of rectangular and triangular sensitive elements (SE) in sensor housing. In order to overcome the above drawback an MMA surfactant thing based on a membrane SE for a more uniform distribution of a load over the surface of the SE was proposed.
Aim. To show the advantages of using AlN as the SE material of a ring wave resonator on SAW.
Materials and methods. The theoretical part of the research was carried out using the finite element method. Mathematical processing was implemented in AutoCAD 2019 and in COMSOL Multiphysics 5.4.
Results. The use of AlN, which acts as the SE material for measuring an acceleration based on SAW was proposed. The proposed solution was compared with existing prototypes based on the use of SiO2 / LiNbO3 membranes, which were characterized by strong anisotropic properties. A 3D model of the SE of a ring wave resonator on surface waves was created. Using computer simulations and COMSOL Multiphysics software, it was shown that the thing was capable to withstand exposures in excess of 10 000 g, and an isotropic AlN sensor overcomed the limitations of both the low sensitivity of SiO2 and the low temperature stability of LiNbO3. AlN demonstrated almost double resistance to irreversible mechanical deformations as compared to SiO2, which, in turn, allows an additional 1.5-fold increase in sensitivity compared to quartz – based sensors.
Conclusion. Based on the data obtained by the modeling, it can be concluded that the use of AIN as SE material is promising, especially for measuring high acceleration values, but with restrictions on temperature sensitivity of the material.
METROLOGY, INFORMATION AND MEASURING DEVICES AND SYSTEMS
Introduction. The exponential growth of measurement information caused by ongoing complication of technical and production facilities necessitates the development of improved or brand new information and measurement systems, including those performing adaptive automatic control functions. Automatic criteria-based selection and reduction of measurement information continuously supplied by multi-parameter sources characterizing the objects under study require algorithms ensuring reconfiguration of automatic control systems during operation. In comparison with automatic control systems based on time-division channelling, the considered adaptive systems provide timely information on the pre-emergency and emergency operation of a facility.
Aim. To develop an algorithmic support for adaptive automatic control systems using asynchronous-cyclic and parallel-sequential operating algorithms, as well as to compare the proposed algorithms in terms of their, control reliability, compression ratio, operation speed and the error associated with multi-channelling.
Materials and methods. The algorithms proposed for supporting the operation of adaptive systems were developed on the basis of queuing theory and simulation modelling using the MatLab/Simulink programming languages, C++.
Results. The developed algorithmic support for automatic control systems based on asynchronous-cyclic analysis of deviations allows the amount of redundant information to be reduced by more than 4 times and the operation speed to be increased by 1.5 times. The developed algorithmic support for automatic control systems based on parallel-sequential analysis of deviations allows the error associated with multi-channelling to be reduced by 1.4 times, thereby bringing the control reliability of such systems closer to that of continuous-control systems. An analysis of the graphs of the error associated with multi-channelling showed that the automatic control systems based on parallel-sequential operational algorithms are invariant to the law of distribution of input quantities, compared to the systems based on asynchronous-cyclic operational algorithms.
Conclusions. The proposed algorithmic support can significantly decrease the redundancy of information and improve the metrological characteristics of automatic control systems. The use of the developed algorithms in automatic control systems based on time-division channelling render their control reliability comparable with that of continuous-control systems.
ISSN 2658-4794 (Online)