Algorithm of semantic image segmentation for solving the problem of positionong an aircraft on the Earth`s surface


DOI: 10.34759/trd-2023-130-18

Аuthors

Olkina D. S.

Moscow Institute of Physics and Technology (National Research University), 9, Institutskiy per., Dolgoprudny, Moscow region, 141701, Russia

e-mail: olkina.ds@phystech.edu

Abstract

This work is devoted to one of the main methodological issues of using the object-oriented approach — choosing the method of segmentation of multichannel images. The methods for determining the approach to the development of a segmentation algorithm are considered, as well as the choice of distinctive features to solve the problem of segmentation of images of the earth’s surface to determine the coordinates of an unmanned aerial vehicle (UAV) in order to position it on the earth’s surface.

The theoretical significance of the study is a numerical study of the formulation of the tasks of segmentation of images by stochastic methods. Of practical importance is obtaining the results of experiments on the segmentation of images of the visible and radio ranges, development of training tools and a software package that solves the problems of segmentation of multichannel images.

The main objectives of the study are to develop a model of neural network segmentation of multichannel images of optical and radio range, сonstruction of the procedure for training segmentation models, obtaining comparative estimates of the computational complexity of algorithms and learning characteristics, hyperparametric optimization of various neural network models, obtaining the main dependencies of training parameters, quality and speed of models. The work uses the neural network of semantic segmentation Bisenet, which allows you to get segmented images of the earth’s surface in real time. For the neural network Bisenet, some hyperparameters were configured: the most suitable optimizer and the best strategy for reducing the learning speed were identified (learning rate).

Keywords:

computer vision, convolutional neural network, segmentation, Bisenet architecture, real-time mode, analogue of inertial navigation

References

  1. Demidova L.A., Nesterov N.I., Tishkin R.V. Telekommunikatsii. Upravlenie, 2014, no. 3 (198), pp. 37-47.
  2. Sheleg S.V. Semanticheskaya segmentatsiya aerofotosnimkov sel’skokhozyaistvennoi rastitel’nosti na baze svertochnykh neironnykh setei (Semantic segmentation of aerial photographs of agricultural vegetation on the basis of convolutional neural networks): autoref. diss. ... Master of Technical Sciences. Minsk, BGUIR, 2021, 9 p.
  3. Alan T’yuring. Vychislitel’nye mashiny i razum (Computers and the Mind), Moscow, Izd-vo AST, 2018, 128 p.
  4. Klifford Pikover. Iskusstvennyi intellekt. Illyustrirovannaya istoriya. Ot avtomatov do neirosetei (Artificial intelligence. Illustrated story. From automata to neural networks), Moscow, Izd-vo Sindbad, 2021, 224 p.
  5. Demidova L.A., Kirakovskii V.V. Trudy nauchno-prakticheskoi konferentsii «Upravlenie sozdaniem i razvitiem sistem, setei i ustroistv telekommunikatsii, Saint-Petersburg, NOTs «Perspektiva», 2008.
  6. Rhee F. C.-H. Uncertain fuzzy clustering: insights and recommendations, IEEE Computational intellegence magazine, 2007, vol. 2, no. 1, pp. 44-56. DOI:10.1109/MCI.2007.357193
  7. Nikolenko S., Kadurin A., Arkhangel’skaya E. Glubokoe obuchenie. Pogruzhenie v mir neironnykh setei (Deep learning. Immersion in the world of neural networks), Saint Petersburg, Izdatel’skii dom «Piter», 2018, 482 p.
  8. Changqian Yu, Changxin Gao, Jingbo Wang, Gang Yu, Chunhua Shen, Nong Sang. BiSeNet V2: Bilateral Network with Guided Aggregation for Real-time Semantic Seg-mentation, International Journal of Computer Vision, 2020, vol. 129 (11), pp. 1-18. DOI:10.1007/s11263-021-01515-2
  9. Yingxiao Xu, Long Pan, Chun Du, Jun Li, Ning Jing, and Jiangjiang Wu. Vision-based UAVs Aerial Image Localization: A Survey, DOI: 10.1145/3281548.3281556
  10. Efimov A.I., Novikov A.I. Komp’yuternaya optika, 2016, vol. 40, no. 2, pp. 258–265.
  11. Ermakov P.G., Gogolev A.A. Trudy MAI, 2021, no. 117. URL: https://trudymai.ru/eng/published.php?ID=156253. DOI: 10.34759/trd-2021-117-11
  12. Ivashova N.D., Mikhailin D.A., Chernyakova M.E., Shanygin S.V. Trudy MAI, 2019, no. 104. URL: https://trudymai.ru/eng/published.php?ID=102223
  13. Sokolov D.Yu. Vooruzhenie i ekonomika, 2012, no. 1 (17), pp. 50-57.
  14. Sokolov D.Yu. Trudy MAI, 2022, no. 123. URL https://trudymai.ru/eng/published.php?ID=165563. DOI: 10.34759/trd-2022-123-17
  15. Guseinov G.A., Zul’fugarly P.R., Abdurrakhmanova I.G. Trudy MAI, 2022, no. 126. URL: https://trudymai.ru/eng/published.php?ID=169012. DOI: 10.34759/trd-2022-126-26
  16. Chigrinets E.G., Verchenko A.V. Trudy MAI, 2019, no. 104. URL: https://trudymai.ru/eng/published.php?ID=102420
  17. Sheinikov A.A., Kovalenko A.M., San’ko A.A. Nauchnyi vestnik MGTU GA, 2023, vol. 26, no. 1, pp. 81-94. URL: https://doi.org/10.26467/2079-0619-202-26-1-81-94
  18. Stepanov D.N., Tishchenko I.P. Programmnye sistemy: teoriya i prilozheniya, 2011, no. 4, pp. 33-43.
  19. Stepanov D.N., Tishchenko I.P. XIV natsional’naya konferentsiya po iskusstvennomu intellektu s mezhdunarodnym uchastiem, KII-2014, Kazan’, Izd-vo RITs «Shkola», 2014, vol. 3, pp. 333-341.
  20. Konushin A. Komp’yuternaya grafika i mul’timedia, 2006, no. 4 (3). URL: http://archive.ec/hwXHz

Download

mai.ru — informational site MAI

Copyright © 2000-2024 by MAI

 Вход