D.E. Boriskin1, N.V. Gorbachev2, I.D. Isaev3, A.N. Savelyev4

1–4 Federal State Autonomous Educational Institution for Higher Education Bauman Moscow State Technical University (national research university) (Moscow, Russia)
1 bde19k023@student.bmstu.ru, 2 gorbachevnv@student.bmstu.ru, 3 isaevid@bmstu.ru, 4 savelyev.an@bmstu.ru

Object detection on digital image frames formed by radar stations is relevant in solving many practical applications. The image quality in different frame fragments is different and depends on both radar stations characteristics and observation conditions.

The effectiveness of CFAR algorithms directly depends on their parameters: the size of analyzed window, guard cells neighboring cell under test (which exclude object’s influence on background parameters assessment), threshold scaling factor. In addition, the choice of CFAR algorithm type (i.e. the method of threshold calculation) also has a significant impact on detection reliability, depending on the nature of surface background and the specifics of detection objects.

Required detection reliability indicators are achieved, as a rule, within the analyzed window, which identifies the need for local adaptation of detection algorithms and their parameters to the specifics of observed scene and observation conditions. Known CFAR algorithms do not always provide adaptation to a variety of detection conditions on the digital radar image frame.

The general structure of adaptive parameter setting procedure for CFAR detection algorithm (sequence and description of basic ope­rations) is presented, the features of training frames forming, calculation of detection reliability indicators, justification for choosing optimal algorithm parameters in analyzed window are described. Digital simulation conditions and results, comparative analysis of CFAR detection algorithms are shown using the example of synthesized frames with random object location for adaptive algorithms with fixed parameters, trained modifications with tabular and regression learning methods, illustrating the advantages of the latter.

Boriskin D.E., Gorbachev N.V., Isaev I.D., Savelyev A.N. Modifications of adaptive object detection algorithms on digital radar image frames // Achievements of modern radioelectronics. 2026. V. 80. № 6. P. 61–71. DOI: https://doi.org/10.18127/j20700784-202606-05

1. Akimov P.S., Bakut P.A., Bogdanovich V.A. i dr. Teoriya obnaruzheniya signalov. Pod red. P.A. Bakuta. M.: Radio i svyaz`. 1984. 440. (in Russian).
2. Akimov P.C., Evstratov F.F., Zaxarov S.I. i dr. Obnaruzhenie radiosignalov. Pod red. A.A. Kolosova. M.: Radio i svyaz`. 1989. 288 s.:il. (in Russian).
3. Levin B.R. Teoreticheskie osnovy` statisticheskoj radiotexniki. M.: Radio i svyaz`. 1989. 656 s. (in Russian).
4. Radioe`lektronny`e sistemy`: Osnovy` postroeniya i teoriya. Spravochnik. Izd. 2-e, pererab. i dop. Pod red. Ya.D. Shirmana. M.: Ra­diotexnika. 2007. 512 s.:il. (in Russian).
5. Informacionny`e texnologii v radiotexnicheskix sistemax: Ucheb. posobie. Pod red. I.B. Fedorova. Izd. 3-e pererab. i dop. M.: Izd-vo MGTU im. N.E`. Baumana. 2011. 846 s.:il. (in Russian).
6. Bakulev P.A. Radiolokacionny`e sistemy`. Uchebnik dlya vuzov. M.: Radiotexnika. 2015. 440 s. (in Russian).
7. Spravochnik po radiolokacii / Pod red M.I. Skolnika; per. s angl. pod red. V.S. Verby`. V 2 knigax. Kn.2. M.: Texnosfera. 2014. (in Russian).
8. Rohling H. Radar CFAR thresholding in clutter and multiple target situations. IEEE Transactions on Aerospace and Electronic Systems. 1983. V. 19. № 4. P. 608–621. doi: 10.1109/taes.1983.309350.
9. Prastitis L.A. On adaptive censored CFAR detection. New Jersey Institute of Technology. 1993. 197 s.
10. Magaz B., Barkat M. Automatic threshold selection in OS-CFAR radar detection using information-theoretic criteria. Progress in Electromagnetics Research. 2014. V. 45. P. 81–94.
11. Shbat M., Tuzlukov V. Evaluation of Detection Performance under Employment of the Generalized Detector in Radar Sensor Systems. Radioengineering. 2014. V. 23. № 1. P. 50–65.
12. Bichler C. Automated Detection of Security-Relevant Objects in Airport Surface Movement Radar. Master’s Thesis – Graz: Graz University of Technology. 2015. 99 r.
13. Monakov A.A. CFAR-obnaruzhitel` celi v radiolokatore s sintezirovannoj aperturoj. Izvestiya vuzov Rossii. Radioe`lektronika. 2024. T. 27. № 3. S. 52–67. (in Russian).
14. Belokurov V.A., Nguen Ch.K. Algoritm stabilizacii urovnya lozhnoj trevogi na fone shuma s nestacionarny`m srednim znacheniem. Izvestiya vuzov Rossii. Radioe`lektronika. 2025. T. 28. № 1. S. 77–87. (in Russian).
15. hang Y., Hao Y. A Survey of SAR Image Target Detection Based on Convolutional Neural Networks. Remote Sensing. 2022. V. 14.
    № 24. Article 6240. DOI: 10.3390/rs14246240.
16. Redmon J., Divvala S., Girshick R., Farhadi A. You Only Look Once: Unified, Real-Time Object Detection. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas. 2016. S. 779–788.
17. Zhao Z.-Q., Zheng P., Xu S.-T., Wu X. Object detection with deep learning: a review. IEEE Transactions on Neural Networks and Learning Systems. 2019. V. 30. № 11. R. 3212–3232.
18. National Institute of Standards and Technology. The NIST Year 2012 Speaker Recognition Evaluation Plan [E`lektronny`j resurs]. Gaithersburg (MD): NIST, 2012. 39 p. URL: https://www.nist.gov/document/nistsre12evalplan-v17-r1pdf (data obrashheniya: 27.05.2025).
19. Lin T.-Y., Goyal P., Girshick R., He K., Dollár P. Focal Loss for Dense Object Detection. Proceedings of the IEEE International Confe­rence on Computer Vision (ICCV, 2017). Venice: 2017. P. 2980–2988. DOI 10.1109/ICCV.2017.324.
20. Magaz B., Barkat M. Automatic threshold selection in OS-CFAR radar detection using information-theoretic criteria. Progress in Electromagnetics Research. 2014. V. 45. P. 81–94.
21. Cover T.M., Hart P.E. Nearest neighbor pattern classification. IEEE Transactions on Information Theory. 1967. V. 13. № 1. P. 21–27.
22. Breiman L. Random Forests. Machine Learning.  2001. V. 45. № 1. S. 5–32.
23. Xasti T., Tibsharani R, Fridman Dzh. Osnovy` statisticheskogo obucheniya: intellektual`ny`j analiz danny`x, prognozirovanie i obnaruzhenie zavisimostej: Ucheb. posobie. M.: ID «Vil`yams». 2017. 744 s. (in Russian).
24. Ho T.K. Random decision forests. Proceedings of the Third International Conference on Document Analysis and Recognition (ICDAR’95). Montreal. QC. 14–16 Aug. 1995. P. 278–282.
25. Bishop C.M. Pattern recognition and machine learning. New York: Springer. 2006. 738 p.