Keypoint-based algorithm for registration of heterogeneous images
DOI: 10.21293/1818-0442-2026-29-2-107-115
Abstract: This paper discusses an algorithm for refining heterogeneous images registration. Image registration is performed using neural network-based selection of candidate regions. Relevance. The registration of heterogeneous images can serve as a basis for unmanned aerial vehicle (UAV) positioning systems, which is a relevant task when signals from global navigation satellite systems are unavailable or distorted due to interference. Purpose of the study. The purpose of this study is to develop and experimentally evaluate an image registration refinement algorithm that combines fast neural network-based selection of candidate regions with precise positioning based on matching and geometric verification of keypoints. Particular attention is paid to scenarios in which a frame acquired by an onboard UAV sensor crosses the boundaries of multiple satellite imagery tiles. Scientific novelty. The scientific novelty lies in the position of a UAV in the absence of signals from global navigation satellite systems by registering heterogeneous images of the underlying surface within a specified flight area. At the first stage, candidate regions are selected using a Siamese convolutional neural network. The network analyzes pairs consisting of a current onboard image and images of the underlying surface, calculating a numerical visual similarity score for each pair, thereby reducing the search area. At the second stage, coordinates are refined based on keypoint matching. Methods. Methods for detecting, describing, and matching local features are considered. As a result of applying the refinement algorithm to the test sample, the average alignment error decreased from 300 pixels after neural network selection to 20 pixels. The most accurate results were obtained for cases in which the UAV frame was entirely located within a single satellite imagery tile. Practical significance. The developed algorithm enables UAV positioning with an accuracy of 10-20 m and a processing time of ≤ 207 ms based on heterogeneous images of the underlying surface.
Keywords: heterogeneous images, image registration, Siamese neural network architecture, keypoints, visual similarity
For citation:
Babaev S. I., Kostrov B. V., Muratov E. R., Tarasov A. S., Tarasova V. Yu. Keypoint-based algorithm for registration of heterogeneous images. Doklady Tomskogo gosudarstvennogo universiteta sistem upravleniya i radioelektroniki, 2026, vol. 29, no. 2, pp. 107–115. DOI: 10.21293/1818-0442-2026-29-2-107-115
Authors and copyright holders:
- Babaev S. I. , Ryazan State Radio Engineering University named after V.F. Utkin (RSREU)
- Kostrov B. V. , Ryazan State Radio Engineering University named after V.F. Utkin (RSREU)
- Muratov E. R. , Ryazan State Radio Engineering University named after V.F. Utkin (RSREU)
- Tarasov A. S. , Ryazan State Radio Engineering University named after V.F. Utkin (RSREU)
- Tarasova V. Yu. , Ryazan State Radio Engineering University named after V.F. Utkin (RSREU)
- 1. Novikov A.I., Efimov A.I., Kolchaev D.A. [Algorithms for automatic identification of objects on heterogeneous images and images superimposition]. Informatsionnye tekhnologii i nanotekhnologii. [Information Technologies and Nanotechnologies] Proc. of IV International Conference and Youth School ITNT–2018, Samara, Novaya tekhnika, 2018, pp. 650–663 (in Russ.).
- 2. Efimov A.I., Kryuchkova T.N., Yaroslavtseva A.I. [Combined Algorithm for Combining Point Clouds]. Proc. of 33rd International Conference on Computer Graphics and Vi- sion GraphiCon 2023, 2023, no. 33, pp. 539–548 (in Russ.).
- 3. Sablina V.A., Efimov A.I., Novikov A.I. Combined ap- proach to object contour superimposition in heterogeneous images. 2017 6th Mediterranean Conference on Embedded Computing (MECO), 2017, pp. 7977175.
- 4. Lebedev M.A., Bondarenko M.A., Komarov D.V., Stepaniants D.G., Vygolov O.V., Vizilter Yu.V., Zheltov S.Yu. [An algorithm for real and synthetic images matching with application to aviation combined vision system]. Vestnik komp'iuternykh i informatsionnykh tekhnologii, 2014, no. 7 (121), pp. 8–13 (in Russ.).
- 5. Vizilter Yu.V., Gorbatsevich V.S., Rubis A.Yu., Vygolov O.V. Image shape matching using diffusion morphology and diffusion correlation. Computer optics, 2015, vol. 39, no. 2, pp. 265–274 (in Russ.).
- 6. Vizilter Yu.V., Vishnyakov B.V., Zheltov S.Yu. Sov- remennye tekhnologii iskusstvennogo intellekta i ikh primenenie v aviatsionnykh kompleksakh [Modern artificial intelligence technologies and their application in aviation systems]. XVI Vserossiyskaya mul'tikonferentsiya po problemam uprav- leniya (MKPU-2023) [Proc. of XVI All-Russian Multiconference on Control Problems], Volgograd, Volgograd State Technical University, 2023, pp. 13–16 (in Russ.).
- 7. Kocheturov V.V., Fedyaev O.I. Siameskie svertochnye neyronnye seti dlya raspoznavaniya lits. [Siamese convolu- tional neural networks for face recognition] Programmnaya in- zheneriya: metody i tekhnologii razrabotki informatsionno- vychislitel'nykh sistem (PIIVS-2024) [Software Engineering: Methods and Technologies for Developing Information and Computing Systems]. Proc. of the 5th International Scientific and Practical Conference, Donetsk, Donetsk National Technical University, 2024, pp. 178–184 (in Russ.).
- 8. Ladanova E.O. [Development of a methodology and a new structure of the Siamese neural network for analysis of ge- ospatial data]. Scientific and Technical Volga region Bulletin, 2024, no. 12, pp. 144–147 (in Russ.).
- 9. Li Y., Chen C.L.Ph., Zhang T. A Survey on Siamese Network: Methodologies, Applications, and Opportunities. IEEE Transactions on Artificial Intelligence, 2022, vol. 3, no. 6, pp. 994–1014. DOI: 10.1109/TAI.2022.3207112.
- 10. Broughton G., Linder P., Rouček T., Vintr T., Krajník T. Robust Image Alignment for Outdoor Teach-and-Repeat Navigation. Proceedings of the 2021 European Conference on Mobile Robots (ECMR), Bonn, Germany, IEEE, 2021, pp. 1–6. DOI: 10.1109/ECMR50962.2021.9568832.
- 11. Rozsypalek Z., Broughton G., Linder P., Roucek T., Kusumam K., Krajnik T. Semi-Supervised Learning for Image Alignment in Teach and Repeat Navigation. Proceedings of the 37th ACM/SIGAPP Symposium on Applied Computing (SAC '22), New York, USA, ACM, art. 4. DOI: 10.1145/3477314.
- 12. Teague S., Chahl J. An Algorithm for Affordable Vi- sion-Based GNSS-Denied Strapdown Celestial Navigation. Drones, 2024, vol. 8, no. 11, art. 652. DOI: 10.3390/drones 8110652.
- 13. Alpatov B.A., Babayan P.V., Koblov Yu.S., Muravyev V.S., Strotov V.V., Feldman A.B. [Software for automatization of scientific investigations aimed at development of perspective vision-based navigation systems for unmanned aerial vehicles]. Vestnik of Ryazan state radio engineering university, 2012, no. 40, pp. 9–15 (in Russ.).
- 14. Sheinikov A.A., Malkin V.A., Smolensky E.A., Ivan- itsky L.A. [Optical-electronic correlation-extreme navigation system of small unmanned aerial vehicle]. System analysis and applied information science, 2024, no. 2, pp. 30–40. DOI: 10.17212/2307-5981-2024-2-30-40 (in Russ.).
- 15. Konnikov E.A., Sergeev D.A., Golikov G.I., Rodionov D.G., Polyakov P.A., Tikhomirov A.F. [TTC-oriented visual navigation systems for UAVS]. Software systems and computational methods, 2025, no. 4, pp. 1–26. DOI: 10.7256/2454- 0714.2025.4.75830 (in Russ.).
- 16. Fortuna-Cervantes J., Ramírez-Torres M., Martinez- Carranza J., Murguía-Barra J., Mejía-Carlos M. Object Detection in Aerial Navigation using Wavelet Transform and Convolutional Neural Networks: A first Approach. Proceedings of the Institute for System Programming of the RAS (Proceedings of ISP RAS), 2021, vol. 33, no. 2, pp. 149–162. DOI: 10.15514/IS- PRAS-2020-33(2)-9 (in Russ.).
- 17. Lowe D.G. Distinctive Image Features from Scale-In- variant Keypoints. International Journal of Computer Vision, 2004, vol. 60, pp. 91–110. DOI: 10.1023/B:VISI.0000029664. 99615.94.
- 18. Denisov A.A., Novikov A.I. [Analysis of image key point detection, description and matching methods]. Vestnik of Ryazan state radio engineering university, 2024, no. 89, pp. 104–116 (in Russ.).
- 19. Bay H., Tinne T., Luc V.G. SURF: Speeded Up Robust Features. Proc. of. Computer Vision – ECCV 2006. ECCV 2006, Lecture Notes in Computer Science, vol. 3951. Springer, Berlin, Heidelberg. https://doi.org/10.1007/11744023_3.
- 20. Mikolajczyk K., Schmid C.A. Performance Evaluation of Local Descriptors. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2005, pp. 1615–1630. DOI: 10.1109/TPAMI.2005.188.
- 21. Dyshlyuk V.O. Issledovanie pokazateley kachestva i bystrodeystviya poiska opornykh tochek na izobrazheniyakh metodom SURF [Investigation of quality and speed metrics for keypoint detection in images using the SURF method]. Molodoy uchenyi, 2018, no. 27, pp. 23–26 (in Russ.).
- 22. DeTone D., Malisiewicz T., Rabinovich A. Super- Point: Self-Supervised Interest Point Detection and Description. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), Salt Lake City, UT, USA, IEEE. DOI: 10.1109/CVPRW.2018.00060.
- 23. Kostrov B.V., Babaev S.I., Efimov A.I., Tarasova V.Yu. [Convolutional neural network hybrid architecture development using spectral transformations]. Izvestiya SFedU. Engi- neering Sciences, 2026, no. 1 (249), pp. 235–249. DOI: 10.18522/2311-3103-2026-1-235-249 (in Russ.).
- 24. Kostrov B.V., Babaev S.I., Tarasova V.Yu. [Issues to the data set formation and preprocessing in the training convo- lutional Siamese neural networks tasks for an aircraft position correcting based on aerial photography]. XXI century: resumes of the past and challenges of the present plus, 2025, vol. 14, no. 3 (71), pp. 70–77 (in Russ.).
- 25. Li J., Hu Q., Ai M. Point Cloud Registration Based on One-Point RANSAC and Scale-Annealing Biweight Estimation. IEEE Transactions on Geoscience and Remote Sensing, 2021, vol. 59, no. 11, pp. 9716–9729. DOI: 10.1109/TGRS. 2020.3048026.
- 26. Hu S., Feng M., Nguyen R.M.H., Lee G.H. CVM-Net: Cross-View Matching Network for Image-Based Ground-to- Aerial Geo-Localization. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2018, Salt Lake City, UT, USA: IEEE, pp. 7258–7267. DOI: 10.1109/CVPR.2018.00758.
- 27. Kostrov B.V., Ustyukov D.I., Tarasova V.Yu. Svertochnaya neyronnaya set' s gibridnymi sloyami dlya opredeleniya vizual'nogo skhodstva sputnikovykh snimkov [Convolutional neural network with hybrid layers for determining the visual similarity of satellite images]. Certificate of State Registration of a Computer Program No. 2025691610, 2025 (in Russ.).
- 28. Nikiforov M.B., Tarasova V.Yu. [Method for deter- mining photo shooting location based on visual search]. XXI century: resumes of the past and challenges of the present plus, 2023, vol. 12, no. 4 (64), pp. 103–110 (in Russ.).
Last issue
Editorial office address
Executive Secretary of the Editor’s Office
Editor’s Office: 40 Lenina Prospect, Tomsk, 634050, Russia
Phone / Fax: + 7 (3822) 701-582
Executive Secretary of the Editor’s Office
Editor’s Office: 40 Lenina Prospect, Tomsk, 634050, Russia
Phone / Fax: + 7 (3822) 51-21-21 / 51-43-02
Subscription for updates
