Mykola P. Romanchuk
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Mykola P. Romanchuk has a total of 40 co-authors. Here are the top co-authors by number of shared publications:
- Olena M. Naumchak
Korolov Zhytomyr Military Institute
5 shared publications - Dmytro L. Fedorchuk
Korolov Zhytomyr Military Institute
4 shared publications - Leonid M. Naumchak
Korolov Zhytomyr Military Institute
4 shared publications - Ihor A. Pilkevych
Korolov Zhytomyr Military Institute
3 shared publications Serhiy V. Kovbasiuk
2 shared publicationsAnastasiia Voznytsia
1 shared publication
Latest Publications
- Open Access
- 21 May 2026
- Published by Academy of Cognitive and Natural Sciences in Journal of Edge Computing
- Vol. 5 (1) , 90-104
Object detection method based on instance segmentation of satellite image obtained in the conditions of cloud cover
Modern achievements in the space industry, combined with the continuous development of remote sensing technologies, form the basis for solving problems in various areas. Medium- and high-resolution satellite imagery often plays a key role in decision-making during crises in hard-to-reach areas. In the process of processing remote sensing data, a significant and still unresolved problem is the reconstruction of clouded images. This article analyses various approaches to cloud removal and data quality improvement. The traditional approaches considered have certain limitations associated with the loss of useful information. Particular attention is paid to deep learning methods, which are gaining popularity in solving cloud removal problems because they produce good results. The article discusses different DNN architectures (convolutional neural networks (CNN), conditional generative adversarial networks (cGAN)) and their modifications, identifies their advantages and disadvantages. A significant advantage of neural networks is their ability to adapt to various conditions and image types. The analysis of the disadvantages of fusing purely optical data led to the conclusion that the best solution to the problem of cloud removal from satellite images is to combine optical and radar data. As a result, the architecture of a model for removing clouds from optical satellite imagery using generative adversarial networks, combined with radar imagery, was developed. The theoretical hypotheses were confirmed by testing the model on the SEN12MS-CR dataset.
- 21 May 2026
- Published by Academy of Cognitive and Natural Sciences in Journal of Edge Computing
- Vol. 5 (1) , 90-104
Open AccessObject detection method based on instance segmentation of satellite image obtained in the conditions of cloud cover
Modern achievements in the space industry, combined with the continuous development of remote sensing technologies, form the basis for solving problems in various areas. Medium- and high-resolution satellite imagery often plays a key role in decision-making during crises in hard-to-reach areas. In the process of processing remote sensing data, a significant and still unresolved problem is the reconstruction of clouded images. This article analyses various approaches to cloud removal and data quality improvement. The traditional approaches considered have certain limitations associated with the loss of useful information. Particular attention is paid to deep learning methods, which are gaining popularity in solving cloud removal problems because they produce good results. The article discusses different DNN architectures (convolutional neural networks (CNN), conditional generative adversarial networks (cGAN)) and their modifications, identifies their advantages and disadvantages. A significant advantage of neural networks is their ability to adapt to various conditions and image types. The analysis of the disadvantages of fusing purely optical data led to the conclusion that the best solution to the problem of cloud removal from satellite images is to combine optical and radar data. As a result, the architecture of a model for removing clouds from optical satellite imagery using generative adversarial networks, combined with radar imagery, was developed. The theoretical hypotheses were confirmed by testing the model on the SEN12MS-CR dataset.
- Open Access
- 15 January 2026
- Published by Private Company Technology Center
- p. 3-29
Scientific and methodological framework for processing heterogeneous data in decision support systems
This section of the study presents a scientific and methodological framework for processing heterogeneous data within decision support systems. The research is grounded in the theory of artificial intelligence, specifically focusing on evolving artificial neural networks, fundamental procedures of genetic algorithms, as well as advanced hybrid bio-inspired algorithms. In the course of the study, the authors propose the following:– a method for processing heterogeneous data in organizational and technical systems;– a method for evaluating the reliability of special-purpose radio communication systems using artificial intelligence theory. The implementation of the proposed scientific and methodological framework enables the following:– reduction of the probability of premature convergence of the metaheuristic algorithm within decision support systems;– maintenance of a balance between convergence speed and diversity of the metaheuristic algorithm during decision-making processes;– consideration of the type of uncertainty and data noise in the metaheuristic algorithm when operating within decision support systems;– accounting for available computational resources of the decision support system;– prioritization of search processes by agents within the swarms of the metaheuristic algorithm;– initialization of swarm individuals with consideration of the type of uncertainty present in the system;– precise training of individuals in metaheuristic algorithms;– execution of both local and global searches considering the level of noise in the data describing the analyzed object;– application as a universal tool for assessing the state of analysis objects through hierarchical object representation;– verification of the reliability of the obtained results;– enhancement of the reliability of object state assessments by constructing object-oriented and relational models of the object’s state with varying levels of hierarchy;– avoidance of the local optimum problem.
- 15 January 2026
- Published by Private Company Technology Center
- p. 3-29
Open AccessScientific and methodological framework for processing heterogeneous data in decision support systems
This section of the study presents a scientific and methodological framework for processing heterogeneous data within decision support systems. The research is grounded in the theory of artificial intelligence, specifically focusing on evolving artificial neural networks, fundamental procedures of genetic algorithms, as well as advanced hybrid bio-inspired algorithms. In the course of the study, the authors propose the following:– a method for processing heterogeneous data in organizational and technical systems;– a method for evaluating the reliability of special-purpose radio communication systems using artificial intelligence theory. The implementation of the proposed scientific and methodological framework enables the following:– reduction of the probability of premature convergence of the metaheuristic algorithm within decision support systems;– maintenance of a balance between convergence speed and diversity of the metaheuristic algorithm during decision-making processes;– consideration of the type of uncertainty and data noise in the metaheuristic algorithm when operating within decision support systems;– accounting for available computational resources of the decision support system;– prioritization of search processes by agents within the swarms of the metaheuristic algorithm;– initialization of swarm individuals with consideration of the type of uncertainty present in the system;– precise training of individuals in metaheuristic algorithms;– execution of both local and global searches considering the level of noise in the data describing the analyzed object;– application as a universal tool for assessing the state of analysis objects through hierarchical object representation;– verification of the reliability of the obtained results;– enhancement of the reliability of object state assessments by constructing object-oriented and relational models of the object’s state with varying levels of hierarchy;– avoidance of the local optimum problem.
- Open Access
- 19 September 2025
- Published by Korolov Zhytomyr Military Institute in PROBLEMS OF CONSTRUCTION, TESTING, APPLICATION AND OPERATION OF COMPLEX INFORMATION SYSTEMS
- Vol. 1 (28 (І)) , 65-76
ERROR ANALYSIS OF AN IMPROVED ALGORITHM FOR CALCULATING THE COORDINATES OF OBJECTS BASED ON AERIAL RECONNAISSANCE DATA FROM AN UNMANNED AERIAL VEHICLE
The calculation of geographical coordinates of objects based on aerial reconnaissance data from unmanned aerial vehicles in the face of enemy electronic warfare is an urgent task, the solution of which can increase the efficiency of the use of unmanned aerial vehicles for military missions. In particular, high-quality processing of aerial reconnaissance data obtained from unmanned aerial vehicles in the absence of signals from the GNSS global positioning system through the use of advanced algorithms and methods will help to improve the accuracy and efficiency of determining the coordinates of ground objects. The article proposes an improved algorithm that allows obtaining the coordinates of objects on flat terrain with high accuracy using aerial reconnaissance materials. To do this, the operator selects four points that are visible both on reconnaissance materials and on reference images that are georeferenced. The selected points allow to calculate the geographical coordinates of any point on the aerial reconnaissance materials. The accuracy of the coordinates will be higher for objects located inside a quadrangle with vertices at the selected points. The marked key points allow for a projective transformation that displays how the pixel coordinates of an object in an UAV image are transformed into its geographic coordinates. To ensure high accuracy of coordinate calculation, key points should be selected around the object, i.e., in such a way that it is located inside a quadrangle with vertices at the key points. As a result of the simulations, the maximum error inside the quadrangle of key points is less than 2 m, and the maximum error outside is about 17 m. The average error inside the quadrangle of key points was slightly more than 0.5 m, and outside - about 1 m. After the simulations, the improved algorithm was tested in field tests. For this purpose, several terrain areas were selected, and the coordinates of the objects located on them were determined. Several wing-type UAVs flew over these areas. The data obtained during the field tests do not differ much from those obtained during the simulations.
- 19 September 2025
- Published by Korolov Zhytomyr Military Institute in PROBLEMS OF CONSTRUCTION, TESTING, APPLICATION AND OPERATION OF COMPLEX INFORMATION SYSTEMS
- Vol. 1 (28 (І)) , 65-76
Open AccessERROR ANALYSIS OF AN IMPROVED ALGORITHM FOR CALCULATING THE COORDINATES OF OBJECTS BASED ON AERIAL RECONNAISSANCE DATA FROM AN UNMANNED AERIAL VEHICLE
The calculation of geographical coordinates of objects based on aerial reconnaissance data from unmanned aerial vehicles in the face of enemy electronic warfare is an urgent task, the solution of which can increase the efficiency of the use of unmanned aerial vehicles for military missions. In particular, high-quality processing of aerial reconnaissance data obtained from unmanned aerial vehicles in the absence of signals from the GNSS global positioning system through the use of advanced algorithms and methods will help to improve the accuracy and efficiency of determining the coordinates of ground objects. The article proposes an improved algorithm that allows obtaining the coordinates of objects on flat terrain with high accuracy using aerial reconnaissance materials. To do this, the operator selects four points that are visible both on reconnaissance materials and on reference images that are georeferenced. The selected points allow to calculate the geographical coordinates of any point on the aerial reconnaissance materials. The accuracy of the coordinates will be higher for objects located inside a quadrangle with vertices at the selected points. The marked key points allow for a projective transformation that displays how the pixel coordinates of an object in an UAV image are transformed into its geographic coordinates. To ensure high accuracy of coordinate calculation, key points should be selected around the object, i.e., in such a way that it is located inside a quadrangle with vertices at the key points. As a result of the simulations, the maximum error inside the quadrangle of key points is less than 2 m, and the maximum error outside is about 17 m. The average error inside the quadrangle of key points was slightly more than 0.5 m, and outside - about 1 m. After the simulations, the improved algorithm was tested in field tests. For this purpose, several terrain areas were selected, and the coordinates of the objects located on them were determined. Several wing-type UAVs flew over these areas. The data obtained during the field tests do not differ much from those obtained during the simulations.
- 14 March 2025
- Published by Academy of Cognitive and Natural Sciences
Improved model for detecting randomly oriented objects on remote sensing images
Object detection in optical remote sensing images is an important task. In recent years, methods based on convolutional neural networks have shown progress. However, due to object variations such as scale, aspect ratio, and random orientation, detection is difficult to further improve. Most convolutional neural networks use rectangular bounding boxes for object detection parallel to the image coordinate axes, which is effective. However, for military objects in satellite images, which may have a large aspect ratio and be randomly oriented, rectangular bounding boxes may not always provide sufficient target localization. In this paper, methods based on the rotation of rectangular frames or other polygonal boundaries are considered, including the following. Rotation Region Proposal Network (RRPN) and Rotation Region CNN (R2CNN). One-stage models such as SSD, YOLO, and RetinaNet have demonstrated high speed and accuracy. The new YOLOv11 model, which is a further development of the one-stage model approaches, demonstrates an increase in the accuracy and speed of object detection and recognition. The purpose of the study is the analysis of modern neural network models and their improvement to enhance the accuracy of detecting and recognizing small densely located, randomly oriented objects on satellite images. The paper proposes a model with a five-parameter regression that includes the parameter of the rotation angle of the bounding box. The results of the study show that this model improves the accuracy of object detection in complex scenarios by providing accurate determination of their orientation and scale.
- 14 March 2025
- Published by Academy of Cognitive and Natural Sciences
Improved model for detecting randomly oriented objects on remote sensing images
Object detection in optical remote sensing images is an important task. In recent years, methods based on convolutional neural networks have shown progress. However, due to object variations such as scale, aspect ratio, and random orientation, detection is difficult to further improve. Most convolutional neural networks use rectangular bounding boxes for object detection parallel to the image coordinate axes, which is effective. However, for military objects in satellite images, which may have a large aspect ratio and be randomly oriented, rectangular bounding boxes may not always provide sufficient target localization. In this paper, methods based on the rotation of rectangular frames or other polygonal boundaries are considered, including the following. Rotation Region Proposal Network (RRPN) and Rotation Region CNN (R2CNN). One-stage models such as SSD, YOLO, and RetinaNet have demonstrated high speed and accuracy. The new YOLOv11 model, which is a further development of the one-stage model approaches, demonstrates an increase in the accuracy and speed of object detection and recognition. The purpose of the study is the analysis of modern neural network models and their improvement to enhance the accuracy of detecting and recognizing small densely located, randomly oriented objects on satellite images. The paper proposes a model with a five-parameter regression that includes the parameter of the rotation angle of the bounding box. The results of the study show that this model improves the accuracy of object detection in complex scenarios by providing accurate determination of their orientation and scale.
- Open Access
- 30 October 2024
- Published by Private Company Technology Center in Eastern-European Journal of Enterprise Technologies
- Vol. 5 (2 (131)) , 6-12
Removing cloudiness on optical space images by a generative adversarial network model using SAR images
The object of this study is the process of removing cloudiness on optical space images. Solving the cloudiness removal task is an important stage in processing data from the Earth remote probing (ERP) aimed at reconstructing the information hidden by these atmospheric disturbances. The analyzed shortcomings in the fusion of purely optical data led to the conclusion that the best solution to the cloudiness removal problem is a combination of optical and radar data. Compared to conventional methods of image processing, neural networks could provide more efficient and better performance indicators due to the ability to adapt to different conditions and types of images. As a result, a generative adversarial network (GAN) model with cyclic-sequential 7-ResNeXt block architecture was constructed for cloud removal in optical space imagery using synthetic aperture radar (SAR) imagery. The model built generates fewer artifacts when transforming the image compared to other models that process multi-temporal images. The experimental results on the SEN12MS-CR data set demonstrate the ability of the constructed model to remove dense clouds from simultaneous Sentinel-2 space images. This is confirmed by the pixel reconstruction of all multispectral channels with an average RMSE value of 2.4 %. To increase the informativeness of the neural network during model training, a SAR image with a C-band signal is used, which has a longer wavelength and thereby provides medium-resolution data about the geometric structure of the Earth's surface. Applying this model could make it possible to improve the situational awareness at all levels of control over the Armed Forces (AF) of Ukraine through the use of current space observations of the Earth from various ERP systems
- 30 October 2024
- Published by Private Company Technology Center in Eastern-European Journal of Enterprise Technologies
- Vol. 5 (2 (131)) , 6-12
Open AccessRemoving cloudiness on optical space images by a generative adversarial network model using SAR images
The object of this study is the process of removing cloudiness on optical space images. Solving the cloudiness removal task is an important stage in processing data from the Earth remote probing (ERP) aimed at reconstructing the information hidden by these atmospheric disturbances. The analyzed shortcomings in the fusion of purely optical data led to the conclusion that the best solution to the cloudiness removal problem is a combination of optical and radar data. Compared to conventional methods of image processing, neural networks could provide more efficient and better performance indicators due to the ability to adapt to different conditions and types of images. As a result, a generative adversarial network (GAN) model with cyclic-sequential 7-ResNeXt block architecture was constructed for cloud removal in optical space imagery using synthetic aperture radar (SAR) imagery. The model built generates fewer artifacts when transforming the image compared to other models that process multi-temporal images. The experimental results on the SEN12MS-CR data set demonstrate the ability of the constructed model to remove dense clouds from simultaneous Sentinel-2 space images. This is confirmed by the pixel reconstruction of all multispectral channels with an average RMSE value of 2.4 %. To increase the informativeness of the neural network during model training, a SAR image with a C-band signal is used, which has a longer wavelength and thereby provides medium-resolution data about the geometric structure of the Earth's surface. Applying this model could make it possible to improve the situational awareness at all levels of control over the Armed Forces (AF) of Ukraine through the use of current space observations of the Earth from various ERP systems
- Open Access
- 16 May 2023
- Published by Academy of Cognitive and Natural Sciences in Journal of Edge Computing
- Vol. 2 (1) , 24-36
Approach to the fake news detection using the graph neural networks
The experience of Russia’s war against Ukraine demonstrates the relevance and necessity of understanding the problems of constant disinformation, the spread of propaganda, and the implementation of destructive negative psychological influence. The issue of dissemination in online media informational messages containing negative psychological influence was researched. Ways of improving the system of monitoring online media using the graph neural networks are considered. The methods of automated fake news detection, based on graph neural networks, were reviewed. The purpose of the article is the analysis of existing approaches that allow identifying destructive signs of influence in text data. It is found that the best way to automate the content analysis process is to use the latest machine learning methods. It was determined and substantiated that graph neural networks are the most reliable and effective solution for the specified task. An approach to automating this procedure based on graph neural networks has been designed and analyzed, which will allow timely and efficient detection and analysis of fake news in the information space of our country. During the research, the process of detecting fake news was simulated. The obtained results showed that the described models of graph neural networks can provide good results in solving the tasks of timely detection and response to threats posed by fake news spread by Russia.
- 16 May 2023
- Published by Academy of Cognitive and Natural Sciences in Journal of Edge Computing
- Vol. 2 (1) , 24-36
Open AccessApproach to the fake news detection using the graph neural networks
The experience of Russia’s war against Ukraine demonstrates the relevance and necessity of understanding the problems of constant disinformation, the spread of propaganda, and the implementation of destructive negative psychological influence. The issue of dissemination in online media informational messages containing negative psychological influence was researched. Ways of improving the system of monitoring online media using the graph neural networks are considered. The methods of automated fake news detection, based on graph neural networks, were reviewed. The purpose of the article is the analysis of existing approaches that allow identifying destructive signs of influence in text data. It is found that the best way to automate the content analysis process is to use the latest machine learning methods. It was determined and substantiated that graph neural networks are the most reliable and effective solution for the specified task. An approach to automating this procedure based on graph neural networks has been designed and analyzed, which will allow timely and efficient detection and analysis of fake news in the information space of our country. During the research, the process of detecting fake news was simulated. The obtained results showed that the described models of graph neural networks can provide good results in solving the tasks of timely detection and response to threats posed by fake news spread by Russia.
- 15 April 2023
- Published by Academy of Cognitive and Natural Sciences
Object detection method based on aerial image instance segmentation received by unmanned aerial vehicles in the conditions rough for visualization
The article analyses the possibilities to use the unmanned aerial complexes in the system of decision making process for the crisis situations that require the object detection at aerial images received by the unmanned aerial vehicle under the conditions of atmospheric fog and smoke over the territories. For image sharpening we used Pansharpening method for injecting the dimensional details from panchromatic image to multispectral image. In order to increase the operational efficiency and accuracy of automotive vehicles detection at aerial images received by the unmanned aerial vehicles for more efficient use of received information in the system of decision making support it was selected Hybrid Task Cascade for Instance Segmentation model. This model is more appropriate for solving the tasks of small-sized object multiclass classification and detection at aerial image using the indirect signs.
- 15 April 2023
- Published by Academy of Cognitive and Natural Sciences
Object detection method based on aerial image instance segmentation received by unmanned aerial vehicles in the conditions rough for visualization
The article analyses the possibilities to use the unmanned aerial complexes in the system of decision making process for the crisis situations that require the object detection at aerial images received by the unmanned aerial vehicle under the conditions of atmospheric fog and smoke over the territories. For image sharpening we used Pansharpening method for injecting the dimensional details from panchromatic image to multispectral image. In order to increase the operational efficiency and accuracy of automotive vehicles detection at aerial images received by the unmanned aerial vehicles for more efficient use of received information in the system of decision making support it was selected Hybrid Task Cascade for Instance Segmentation model. This model is more appropriate for solving the tasks of small-sized object multiclass classification and detection at aerial image using the indirect signs.
- 15 April 2023
- Published by Academy of Cognitive and Natural Sciences
An analysis of approach to the fake news assessment based on the graph neural networks
The experience of Russia’s war against Ukraine demonstrates the relevance and necessity of under- standing the problems of constant disinformation, the spread of propaganda, and the implementation of destructive negative psychological influence. The issue of dissemination in online media informational messages containing negative psychological influence was researched. Ways of improving the system of monitoring online media using the graph neural networks are considered. The methods of automated fake news detection, based on graph neural networks, were reviewed. The purpose of the article is the analysis of existing approaches that allow identifying destructive signs of influence in text data. It is found that the best way to automate the content analysis process is to use the latest machine learning methods. It was determined and substantiated that graph neural networks are the most reliable and effective solution for the specified task. An approach to automating this procedure based on graph neural networks has been designed and analyzed, which will allow timely and efficient detection and analysis of fake news in the information space of our country. During the research, the process of detecting fake news was simulated. The obtained results showed that the described models of graph neural networks can provide good results in solving the tasks of timely detection and response to threats posed by fake news spread by Russia.
- 15 April 2023
- Published by Academy of Cognitive and Natural Sciences
An analysis of approach to the fake news assessment based on the graph neural networks
The experience of Russia’s war against Ukraine demonstrates the relevance and necessity of under- standing the problems of constant disinformation, the spread of propaganda, and the implementation of destructive negative psychological influence. The issue of dissemination in online media informational messages containing negative psychological influence was researched. Ways of improving the system of monitoring online media using the graph neural networks are considered. The methods of automated fake news detection, based on graph neural networks, were reviewed. The purpose of the article is the analysis of existing approaches that allow identifying destructive signs of influence in text data. It is found that the best way to automate the content analysis process is to use the latest machine learning methods. It was determined and substantiated that graph neural networks are the most reliable and effective solution for the specified task. An approach to automating this procedure based on graph neural networks has been designed and analyzed, which will allow timely and efficient detection and analysis of fake news in the information space of our country. During the research, the process of detecting fake news was simulated. The obtained results showed that the described models of graph neural networks can provide good results in solving the tasks of timely detection and response to threats posed by fake news spread by Russia.