Mathematical model for aerial image decoding and automated analysis of urban infrastructure

• Datum: 2025
• 1. Verfasser: Marushchak, Vasyl
• Format: Artikel
• Sprache: Ukrainian
• Veröffentlicht: Kyiv National University of Construction and Architecture 2025
• Schlagworte: information technologies; image interpretation; aerial photography; geographic information systems; object recognition; aerial data processing; machine learning
• Online Zugang: https://es-journal.in.ua/article/view/335838
• Назва журналу: Environmental safety and natural resources

This study addresses a scientific and practical challenge focused on improving the monitoring of urbanized areas by utilizing remotely piloted aerial systems (RPAS) to enhance the efficiency of detecting urban infrastructure objects (UIOs). The key factors influencing the accuracy and timeliness of UIO identification based on aerial imagery have been analyzed. The study formulates the main requirements for the image interpretation process, emphasizing the need to ensure high levels of detection speed and reliability.Particular attention is paid to the prospects of automating the recognition of UIOs using computer vision technologies and deep learning methods based on artificial neural networks. Within the framework of the research, a conceptual mathematical model for automated image interpretation is proposed, which provides for the detection of urban infrastructure objects through computer vision algorithms implemented via artificial intelligence technologies.Integration of the developed model into the image processing system makes it possible to significantly improve the operational efficiency of UIO recognition while maintaining an acceptable level of interpretation accuracy. The application of such solutions is particularly relevant for real-time monitoring of changes in urban development, the condition of transportation infrastructure, engineering networks, and technogenic facilities. The use of machine learning algorithms not only facilitates object detection but also enables their functional classification with a high degree of accuracy.Implementation of the proposed approach creates prerequisites for the automated processing of large volumes of visual data, ensuring timely responses to changes in the urban environment.Prospects for further work include: accumulation and structuring of a specialized database of aerial photographs with marked objects of urban infrastructure, creation of a comprehensive method for automating the decoding of aerial photographs with its subsequent software implementation, implementation of comprehensive experimental studies aimed at verifying the effectiveness of the proposed mathematical model according to the criteria of efficiency and reliability.