A.I. Sukachev1, M.S. Ryabykh2, E.A. Sukacheva3, N.D. Dolgikh4

1-4 FSBEI of HE “Voronezh State Technical University” (Voronezh, Russia)

1 mag.dip@yandex.ru; 2 Ryabichmaxim@yandex.ru; 3 elena_s_1331@mail.ru; 4 mag.dip@yandex.ru

This paper presents approaches to solving the problem of optical object detection using computer vision methods. The relevance of the study is determined by the need to increase the efficiency and automation of the detection process in conditions where objects are located at a significant distance from the observation point, making their recognition by the operator impossible. The complexity is that the sizes of objects in the image may have different parameters depending on the distance from the observation point, which requires the use of adaptive processing algorithms. As a solution to the problem, an algorithm for complex image preprocessing aimed at extracting scale-invariant features for subsequent detection is proposed. A description of an algorithm for selecting an adaptive brightness threshold using the Otsu method with the maximization of interclass variance is compiled. A mathematical description of the Otsu method is provided, describing the threshold selection methodology. An algorithm for obtaining a binary image edge mask using Gaussian smoothing and hysteresis threshold filtering using the Canny edge detector method is described. This paper examines the Hough transform, which allows one to determine the location of a detected object in an image based on a priori information about the object's patterns. A practical solution to the problem of image preprocessing using a software method is presented, including constructing a binary boundary mask and indicating the detected object in a test image. The practical significance of the work is confirmed by implementing the proposed approach using software tools. Using the software package, indication of a detected object in a test image is demonstrated. The obtained results can be used in monitoring systems for detecting remote objects. A further direction for research will be the integration of machine learning methods into the developed algorithm for classifying detected objects by type.

This work was supported by the Ministry of Science and Higher Education of the Russian Federation (project no. FZGM-2025-0002)

Sukachev A.I., Ryabykh M.S., Sukacheva E.A., Dolgikh N.D. Passive optical detection method for unmanned aerial vehicles at the maximum registration range // Radiotekhnika. 2026. V. 90. № 7. P. 88−92. DOI: https://doi.org/10.18127/j00338486-202607-16

1. Boguslavskij A.A., Sokolov S.M. Obnaruzhenie obrazov ob"ektov s izmenyayushchejsya strukturoj v sistemah tekhnicheskogo zreniya real'nogo vremeni. Informacionnye tekhnologii i vychislitel'nye sistemy. 2006. № 1. S. 61–77 (in Russian).
2. Astapova M.A., Lebedev I.V. Obzor podhodov k detektirovaniyu defektov elementov LEP na izobrazheniyah v infrakrasnom, ul'trafioletovom i vidimom spektrah. Modelirovanie, optimizaciya i informacionnye tekhnologii. 2020. T. 8. № 4(31). S. 36-69.
   DOI: 10.26102/2310-6018/2020.31.4.036 (in Russian).
3. OpenCV: Introduction [Elektronnyj resurs]. URL: https://docs.opencv.org/4.x/d1/dfb/intro.html.
4. Szeliski R. Computer vision: algorithms and applications. 2nd ed. Springer. 2022. 925 p.
5. Otsu N.A. Threshold selection method from gray-level histograms. IEEE Transactions on Systems, Man, and Cybernetics. 1979. V. 9. № 1. P. 62–66.
6. Forsyth D.A., Ponce J. Computer vision: a modern approach. 2nd ed. Pearson. 2011. 792 p.