Sualtı Nesnelerinin Sınıflandırılmasında Zaman Serisi Görüntü Dönüşümü Yöntemlerinin Yeni Bir Yaklaşımı ve Uygulaması

Abstract

Sonar, ses dalgalarını kullanarak boyutu, uzaklığı, yönü ve diğer nesne özelliklerini belirlemek için kullanılır. Denizaltı madenciliği, petrol arama, denizaltı haritalama, balık sürülerinin takibi ve mayın tespitinde yaygın olarak kullanılmaktadır. Makine Öğrenimi araştırmasında, sonar sinyallerini tanımlamak ve sınıflandırmak için kullanılması gereken özellik çıkarma, seçme, algoritma seçimi ve hiper parametre optimizasyonu, uzun yıllardır çalışılan bilimsel problemler olarak görülmektedir. Bu çalışmada, yaygın olarak kullanılan makine öğrenimi algoritmaları ve öznitelik çıkarma süreçleri yerine, su altı nesnelerini yenilikçi bir yaklaşım olarak sınıflandırmak için üç farklı matematiksel dönüşüm önerilmiştir. Zaman serisi formatında bir veri setine uygulanan bu yeni yaklaşım, veriler tek boyutlu verilerden iki boyutlu bir formata dönüştürülmüş ve bu görüntüleri birleştiren yeni bir görüntü oluşturmak için basit bir kanal birleştirme tekniği uygulanmıştır. Yöntemlerin performansı, sonar veri setinde derin öğrenme algoritmaları kullanılarak mayın ve kayaların sınıflandırma sonuçlarıyla ölçülmüştür. Ayrıca klasik algoritmalar ile ve derin öğrenme ile elde edilen performans sonuçları karşılaştırılmıştır. Son olarak, literatürdeki diğer çalışmalarla karşılaşıldığında, önerilen zaman serisi verilerinden görüntüye dönüşümün kanal birleştirme yaklaşımı ile öznitelik çıkarma ihtiyacını ortadan kaldırdığı ve diğerlerine göre üstün sonuçlar elde ettiği görülmüştür.

Keywords

• MTF
• RP
• GAF
• Sonar
• Derin Öğrenme

A Novel Approach and Application of Time Series to Image Transformation Methods on Classification of Underwater Objects

Abstract

Sonar is used to determine the size, distance, direction, and other object features using sound waves. It is widely used in submarine mining, oil exploration, submarine mapping, tracking fish shoals, and mine detection. In Machine Learning (ML) research, feature extraction, selection, algorithm selection, and hyper-parameter optimization, which should be used to identify and classify sonar signals, are seen as scientific problems studied for many years. In this study, instead of commonly used ML algorithms and feature extraction processes, three different mathematical transformations were suggested to classify the underwater objects as an innovative approach. This novel approach applied on a data set in time-series format, data has been transformed from one-dimensional data to a two-dimensional format and a simple channel merging technique was applied to create a new image joining these images. The methods' performance was measured by the classification results of mines and rocks using deep learning algorithms on the sonar dataset. Moreover, the performance results obtained with deep learning, compared with the classical algorithms. Finally, confronted with other studies in the literature, it has been seen that the proposed time-series data-to-image transformation with a channel-merging approach eliminates the need for feature extraction and achieves superior results against the others.

Keywords

• MTF
• RP
• GAF
• Sonar
• Deep Learning

References

1. [1] A.D. Waite Sonar for Practising Engineers. Wiley, West Sussex, England, 2002.
2. [2] L. Jing “The principle of side-scan sonar and its application in the detection of suspended submarine pipeline treatment,” Materials Science and Engineering, {IOP}, 439, 2018. Doi: 10.1088/1757-899X/439/3/032068.
3. [3] V.L. Lucieer. “Object-oriented classification of side-scan sonar data for mapping benthic marine habitats,” International Journal of Remote Sensing. vol. 29(3), pp. 905–921, 2018. Doi: 10.1080/01431160701311309.
4. [4] A. Burguera, G. Oliver. “High-Resolution Underwater Mapping Using Side-Scan Sonar,” PLOS ONE. vol. 11(1), 2016. Doi: 10.1371/journal.pone.0146396.
5. [5] H.J. Flowers, J.E. Hightower. “A Novel Approach to Surveying Sturgeon Using Side-Scan Sonar and Occupancy Modeling,” Marine and Coastal Fisheries. vol. 5(1), pp. 211–223, 2013. Doi: 10.1080/19425120.2013.816396
6. [6] A.T. Çelebi, M.K. Güllü, S. Ertürk. “Mine detection in side scan sonar images using Markov Random Fields with brightness compensation,” 2011 IEEE 19th Signal Processing and Communications Applications Conference (SIU), 20-22 April 2011, Antalya, Turkey, 916–919, 2011. Available: https: //ieeexplore.ieee.org/document/52929801. [Accessed: 15 Jan. 2021].
7. [7] C.S. Huebner. “Evaluation of side-scan sonar performance for the detection of naval mines.” Target and Background Signatures IV, vol. 10794. SPIE, pp. 158–166, 2018. Doi: https://doi.org/10.1117/12.2325642
8. [8] M. Verleysen, D. François. “The Curse of Dimensionality in Data Mining and Time Series Prediction,” Lecture Notes in Computer Science; vol. 3512(06), pp. 758–770, 2005. Doi : 10.1007/11494669_93

9. [9] R.P. Gorman, T.J. Sejnowski. “Analysis of Hidden Units in a Layered Network Trained to Classify Sonar Targets,” Neural Networks. vol. 1(1), pp. 75–89, 1988. Doi: https://doi.org/10.1016/0893-6080(88)90023-8
10. [10] B. Erkmen, T. Yıldırım. “Improving classification performance of sonar targets by applying general regression neural network with PCA,” Expert Systems with Applications. vol. 35(1-2), pp. 472–475, 2008. Doi: https://doi.org/10.1016/j.eswa.2007.07.021
11. [11] M. Hossin, F. Mahudin, I. Din, A.R. Mat. “Analysis of Nine Instance-Based Genetic Algorithm Classifiers Using Small Datasets,” Journal of Telecommunication, Electronic and Computer Engineering, vol. 9, pp 3–11, 2017.
12. [12] J. Novakovic. “Using Information Gain Attribute Evaluation to Classify Sonar Targets,” 17th Telecommunications forum TELFOR 2009. 24-26 November 2009. Serbia, Belgrade. Available: http://2009.telfor.rs/files/radovi/10_60.pdf
13. [13] S. Fong, S. Deb, R. Wong, et al. “Underwater Sonar Signals Recognition by Incremental Data Stream Mining with Conflict Analysis,” International Journal of Distributed Sensor Networks. vol. 10(5), pp 1-12 May 2014. Doi: 10.1155/2014/635834
14. [14] X. Hong, J. Zhang, S.U. Guan. “Incremental Maximum Gaussian Mixture Partition For Classification,” 2nd Joint International Information Technology, Mechanical and Electronic Engineering Conference (JIMEC 2017); 62, 2017.
15. [15] T. Shang, X. Xia, J. Zheng. “MIME-KNN: Improve KNN Classifier Performance Include Classification Accuracy and Time Consumption.” DEStech Transactions on Computer Science and Engineering. July 2018. Doi: 10.12783/dtcse/csse2018/24490
16. [16] Z. Wang, T. Oates. “Imaging Time-Series to Improve Classification and Imputation.” Proceedings of the Twenty-Fourth International Joint Conference on Artificial Intelligence. (IJCAI), 2015. Available: https://arxiv.org/abs/1506.003https://arxiv.org/abs/1506.00327
17. [17] Z. Wang, T. Oates. “Spatially Encoding Temporal Correlations to Classify Temporal Data Using Convolutional Neural Networks.” Journal of Computer and Systems Sciences 2015.
18. [18] J.P. Eckmann, S.O. Kamphorst, D. Ruelle “Recurrence Plots of Dynamical Systems.” Euro- Physics Letters (EPL). vol. 4(9), pp. 973–977, 1987.
19. [19] A. Şeker, B. Diri, H.H. “Balık. Derin Öğrenme Yöntemleri ve Uygulamaları Hakkında Bir İnceleme.” Gazi Mühendislik Bilimleri Dergisi. vol. 3(3), pp. 47-64, 2017.
20. [20]S. Raschka, V. Mirjalili. Python Machine Learning, 2nd Edition 2017.
21. [21] A. Civrizoğlu Buz, M.U. Demirezen, U. Yavanoğlu. “A Novel Approach and Application of Time Series to Image Transformation Methods on Classification of Underwater Objects,” in Proceedings of the 4th International Conference on Engineering Technologies (ICENTE'20). 19-21 November 2020, Konya, Turkey [Online]. Available: https://icente.selcuk.edu.tr/uploads/files2/ICENTE20_ProceedingsBook_v1.pdf. [Accessed: 15 Jan. 2021].