Informatics and Applications

2024, Volume 18, Issue 4, pp 77-85

NEURAL QUADTREE AND ITS APPLICATIONS FOR SAR IMAGERY SEGMENTATION

A. M. Dostovalova

Abstract

The paper considers the neural ensemble neural network architecture that uses a quadtree model for SAR (Synthetic Aperture Radar) image segmentation under the lack of training data. The Neural Quadtree network (NQN) consists of segmentation network forming the image pixels features and a graph convolution network with the special branch pruning block establishing the spatial and hierarchical connections between pixels. The NQN is used for segmenting of several SAR images that differ a lot both in presented surfaces and characteristics (Sentinel-1, ESAR (Experimental SAR), HRSID (High-Resolution SAR Images Dataset)). A comparison was made of the results of processing images of NQN and a conventional quad-tree using a common U-Net network segmentor. The NQN demonstrates the higher quality in target detection in comparison with a conventional quadtree. The difference in Recall values for such objects classes between NQN and quadtree ranges from 2.13% to 11.63%.

[+] References (28)

Clabaut, E., S. Foucher, Y. Bouroubi. and M. Germain. 2024. Synthetic data for Sentinel-2 semantic segmentation. Remote Sens. — Basel 16(5):818. doi: 10.3390/ rs16050818.
Aleiassaee, A., A. Kumar, R.M. Anwer, et al. 2023. Transformers in remote sensing: A survey. Remote Sens.—Basel 15(7):1860. doi: 10.3390/rs15071860.
Sun, Z., H. Geng, Z. Lu, R. Scherer, and M. Wozniak. 2021. Review of road segmentation for SAR images. Remote Sens.— Basel 13(5):1011. doi: 10.3390/rs13051011.
Ciecholewski, M. 2024. Review of segmentation methods for coastline detection in SAR images. Arch. Comput. Method. E. 31: 839–869. doi: 10.1007/s11831-023-10000-7.
Prokhorets, I.O., and A. S. Stepanov. 2024. Kartografirovanie zemel’ sel’skokhozyaystvennogo naznacheniya Khabarovskogo kraya metodami mashinnogo obucheniya s ispol’zovaniem izobrazheniy [Mapping of the Khabarovsk Region arable lands by machine learning using Sentinel-2 images]. Sistemy i Sredstva Informatiki — Systems and Means of Informatics 34(1):57–69. doi: 10.14357/08696527240105. EDN: YEZNXD.
Song, Y., T. Wang, P. Cai, S.K. Mondal, and J. Sahoo. 2023. A comprehensive survey of few-shot learning: Evolution, applications, challenges, and opportunities. ACM Comput. Surv. 55(13s):271. 40 p. doi: 10.1145/3582688.
Karniadakis, G.E., I.G. Kevrekidis, L. Lu, P. Perdikaris, S. Wang, and L. Yang. 2021. Physics-informed machine learning. Nature Reviews Physics 3:422–440. doi: 10.1038/s42254-021-00314-5.
Gorshenin, A.K., and V. Yu. Kuzmin. 2020. Analiz konfiguratsiy LSTM-setey dlya postroeniya srednesrochnykh vektornykh prognozov [Analisys of configurations of LSTM networks for medium-term vector forecasting]. Informatika i ee primeneniya—Inform. Appl. 14(1):10–16. doi: 10.14357/19922264200102. EDN: SZNTXE.
Gorshenin, A. K., and V. Yu. Kuzmin. 2021. Metod povysheniya tochnosti neyrosetevykh prognozov s ispol’zovaniem smeshannykh veroyatnostnykh modeley i ego realizatsiya v vide tsifrovogo servisa [Method for improving accuracy of neural network forecasts based on probability mixture models and its implementation as a digital service]. Informatika i ee primeneniya — Inform. Appl. 15(3):63–74. doi: 10.14357/19922264210309.EDN: FDRKZX.
Gorshenin, A. K., and V. Yu. Kuzmin. 2022. Statistical feature construction for forecasting accuracy increase and its applications in neural network based analysis. Mathematics 10:589. doi: 10.3390/math10040589.
Gorshenin, A. K., and A. L. Vilyaev. 2022. Finite normal mixture models for the ensemble learning of recurrent neural networks with applications to currency pairs. Pattern Recognition Image Analysis 32:780–792. doi: 10.1134/S1054661822040058.
Gorshenin, A. K., and V. Yu Korolev. 2020. Approksimatsiya raspredeleniy razmerov chastits lunnogo regolita na osnove metoda statisticheskoy simulyatsii vyborok [Approximation of particle size distributions of lunar regolith based on the resampling]. Informatika i ee primeneniya — Inform. Appl. 14(2):50–57. doi: 10.14357/ 19922264200207. EDN: PQOYUJ.
Gorshenin, A. K., S. A. Gorbunov, and D. Yu. Volkanov. 2023. O klasterizatsii ob”ektov setevoy vychislitel’noy infrastruktury na osnove analiza statisticheskikh anomaliy v trafike [Toward clustering of network computing infrastructure objects based on analisys of statistical anomalies in network traffic]. Informatika i ee primeneniya — Inform. Appl. 17(3):76–87. doi: 10.14357/19922264230311. EDN: XHTMVI.
Pastorino, M., A. Montaldo, L. Fronda, et al. 2021. Multisensor and multiresolution remote sensing image classification through a causal hierarchical Markov framework and decision tree ensembles. Remote Sens. — Basel 13(5):849. doi: 10.3390/rs13050849.
Pastorino, M., G. Moser, S.B. Serpico, and J. Zerubia. 2022. Semantic segmentation of remote-sensing images throughfully convolutional neural networks and hierarchical probabilistic graphical models. IEEE T. Geosci. Remote 60:5407116. 16 p. doi: 10.1109/TGRS.2022.3141996.
Dostovalova, A.M. 2023. Using a model of a spatial– hierarchical quadtree with truncated branches to improve the accuracy of image classification. Izv. Atmos. Ocean. Phy. 59(10):1255–1262. doi: 10.1134/S0001433823120071.
Jayaraman, P.K., J. Mei, J. Cai, and J. Zheng. 2018. Quadtree convolutional neural networks. Computer vision. Eds. V. Ferrari,M.Hebert,C. Sminchisescu, and Y.Weiss. Lecture notes in computer science ser. Cham: Springer. 11210:554–569. doi: 10.1007/978-3-030-01231-1 34.
Jewsbury, R., A. Bhalerao, and N.M. Rajpoot. 2021. A quadtree image representation for computational pathology. IEEE/CVF Conference (International) on Computer Vision Science Proceedings. Piscataway, NJ: IEEE. 648–656. doi: 10.1109/ICCVW54120.2021.00078.
Ronen, T., O. Levy, and A. Golbert. 2023. Vision transformers with mixed-resolution tokenization. IEEE/CVF Conference on Computer Vision and Pattern Recognition Proceedings. Piscataway, NJ: IEEE. 4612–4621.
Ke, L., M. Danelljan, X. Li, Y.W. Tai, C.K. Tang, and F. Yu. 2022. Mask transfiner for high-quality instance segmentation. IEEE/CVF Conference on Computer Vision and Pattern Recognition Proceedings. Piscataway, NJ: IEEE. 4402–4411. doi: 10.1109/CVPR52688.2022.00437.
Tang, S., J. Zhang, S. Zhu, and P. Tan. 2023. Quadtree attention for vision transformers. Cornell University. 16 p. Available at: https://arxiv.org/abs/2201.02767 (accessed March 13,2024).
Chitta, K., J.M. Alvarez, and M. Hebert. 2020. Quadtree generating networks: Efficient hierarchical scene parsing with sparse convolutions. IEEE Winter Conference on Applications of Computer Vision Proceedings. Piscataway, NJ: IEEE. 2009–2018. doi: 10.1109/WACV45572.2020.9093449.
Zhang, S., H. Tong, J. Xu, et al. 2019. Graph convolutional networks: A comprehensive review. Computational Social Networks 6(11):94–104. doi: 10.1186/s40649-019-0069-y.
Gorshenin, A. K., and E. I. Guseynova. 2023. Povyshenie dokhodnosti torgovli na FOREX s pomoshch’yu LSTM-identifikatsii svechnykh patternov i indikatora tikovykh ob”emov [Increasing FOREX trading profitability with LSTM candlestick pattern recognition and tick volume indicator]. Informatika i ee primeneniya — Inform. Appl. 16(3): 26–38. doi: 10.14357/19922264220304. EDN: QOUMZY.
Potin, P., P. Bargellini, H. Laur, B. Rosich, and S. Schmuck. 2012. Sentinel-1 mission operations concept. Geoscience and Remote Sensing Symposium (International) Proceedings. Piscataway, NJ: IEEE. 1745–1749. doi: 10.1109/IGARSS.2012.6351183.
Scheiber, R., I. Hajnek, R. Horn, K. P. Papathanassiou, P. Prats, and A. Moreira. 2008. Recent developments and applications of multi-pass airborne interferometric SAR using the E-SAR system. 7th European Conference on Synthetic Aperture Radar Proceedings. Friedrichshafen, Germany: IEEE. 4 p.
Wei, Sh., X. Zeng, Q. Que, M. Wang, and H. Su. 2020. HRSID: A high-resolution SAR images dataset for ship detection and instance segmentation. IEEE Access 8:120234–120254. doi: 10.1109/ACCESS.2020.3005861.
Bergstra, J., and Y. Bengio. 2012. Random search for hyper-parameter optimization. J. Mach. Learn. Res. 13(10): 281–305.

[+] About this article