Journals / Harita Dergisi / 2010 / Cilt: 76 - Sayı: 144

(Investigation of the effects of kernel functions in satellite image classification using support vector machines)

Destek vektör makineleri ile uydu görüntülerinin sınıflandırılmasında kernel fonksiyonlarının etkilerinin incelenmesi

Pages
73–82
DOI
—
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

On the other hand, it is found that all SVM models produced more accurate results compared to maximum likelihood classification.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:Remotely sensed images and the thematic maps derived from these images provide invaluable information about the nature of the Earth surface materials and objects.
📊
Key Findings & Data:In this study, a detailed performance analysis was carried out for SVMs with regard to kernel function and chosen parameter values.
💡
Conclusion & Impact:On the other hand, it is found that all SVM models produced more accurate results compared to maximum likelihood classification.
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:Domain Parameters

Abstract

Remotely sensed images and the thematic maps derived from these images provide invaluable information about the nature of the Earth surface materials and objects. The robustness of classification techniques used to produce these thematic maps can be crucial especially for complex classification problems. Support vector machines (SVMs) are machine learning algorithms that have been recently introduced and successfully used in many fields. Their use in remote sensing is relatively new. It is proved that these algorithms are more robust and effective than other classifiers, particularly the ones on statistical basis. In this study, a detailed performance analysis was carried out for SVMs with regard to kernel function and chosen parameter values. In the analyses four particular kernel functions that have been most widely used in the literature were investigated with their effects on classification accuracy. It is found that radial basis function and Pearson VII function kernels produced the highest performance (>94%) for the data set considered in this study. Also, normalized polynomial kernel showed a poor performance with the lowest classification accuracy (91,78%) among all SVM models. On the other hand, it is found that all SVM models produced more accurate results compared to maximum likelihood classification.

Özet

Uzaktan algılanmış görüntüler ve bu görüntülerin sınıflandırılmasıyla üretilen tematik haritalar, dünya yüzeyindeki objelerin yapısı ve özellikleri hakkında çok önemli bilgiler sunarlar. Tematik haritaların elde edilmesinde kullanılan sınıflandırıcıların etkinliği özellikle zor ve karmaşık sınıflandırma problemleri için büyük önem taşımaktadır. Destek vektör makineleri (DVM) son yıllarda ortaya atılmış ve birçok alanda başarıyla uygulanmış bir makine öğrenme algoritmasıdır. Uzaktan algılama alanındaki uygulamaları ise güncel sayılabilecek niteliktedir. DVM’lerin diğer sınıflandırıcılardan özellikle istatistiksel tabanlı sınıflandırıcılardan daha güçlü ve etkili olduğu ispatlanmıştır. Bu çalışmada, DVM’lerin sınıflandırma performansının kernel fonksiyonu ve parametre değerlerine bağlı değişimi detaylı şekilde analiz edilmiştir. Analizler sırasında literatürde en yaygın kullanılan dört kernel fonksiyonun sınıflandırma doğruluğu üzerindeki etkileri incelenmiştir. Kullanılan veri seti için radyal tabanlı fonksiyon ve Pearson VII fonksiyon kernellerinin en yüksek performansı (>%94 genel doğruluk) gösterdikleri belirlenmiştir. Normalleştirilmiş polinom kerneli ise en düşük sınıflandırma doğruluğunu (%91,78) üreten DVM modeli olmuştur. Diğer taraftan, en çok benzerlik yöntemiyle performans karşılaştırması yapıldığında, tüm DVM modellerinin daha yüksek doğruluklu sonuçlar ürettiği saptanmıştır.

🔖

Cite This Article

KAVZOĞLU, T., ÇÖLKESEN, İ. (2010). (Investigation of the effects of kernel functions in satellite image classification using support vector machines). Harita Dergisi, Cilt: 76 - Sayı: 144, 73–82. https://ajindex.com/articles/186032

References (17)

  1. Arnulf B. A. Graf, S. Borer, 2001, Normalization in Support Vector Machines, Lecture Notes in Computer Science, 2191: 277-282.
  2. Bruzzone, L., Conese, C., Maselli, F., Roli, F., 1997, Multisource Classification of Complex Rural Areas by Statistical and Neural Network Approaches, Photogrammetric Engineering and Remote Sensing, 63(5): 523-533.
  3. Campbell, J.B., 1996,
  4. Introduction to Remote Sensing, Guilford Press,
  5. New York. Congalton R.G., Green, K., 2009, Assessing the Accuracy of Remotely Sensed Data, 2. Baskı,
  6. CRC Press,
  7. New York. Cortes, C., Vapnik, V., 1995,
  8. Support-Vector Network, Machine Learning, 20(3): 273–297.
  9. Erbek, F.S., Özkan, C., Taberner, M., 2003, Comparison of Maximum Likelihood Classification Method with Supervised Artificial Neural Network Algorithms for Land Use Activities, International Journal of Remote Sensing, 25(9): 1733-1748.
  10. Foody, G.M., 1995, Using Prior Knowledge in Artificial Neural Network Classification with a Minimal Training Set. International Journal of Remote Sensing, 16(2): 301-312.
  11. Foody, G.M., Mathur, A., 2004, A Relative Evaluation of Multiclass Image Classification by Support Vector Machines, IEEE Transactions on Geoscience and Remote Sensing, 42(6): 1335–1343.
  12. Huang, C., Davis, L.S., Townshed, J.R.G., 2002, An Assessment of Support Vector Machines for Land Cover Classification, International Journal of Remote Sensing, 23(4), 725–749.
  13. Hsu, C.W., Chang, C.C., Lin, C.J., 2010, A Practical Guide to Support Vector Classification,http://www.csie.ntu.edu.tw/~cjli n/papers/guide/guide.pdf. Kavzoglu, T., Colkesen, I., 2009, A Kernel Functions Analysis for Support Vector Machines for Land Cover Classification, International Journal of Applied Earth Observation and Geoinformation, 11(5): 352- 359. Kavzoglu, T., Mather, P.M., 2003, The Use of Backpropagating Artificial Neural Networks in Land Cover Classification, International Journal of Remote Sensing, 24(23): 4907-4938.
  14. Kavzoglu, T., Reis, S., 2008, Performance Analysis of Maximum Likelihood and Artificial Neural Network Classifiers for Training Sets with Mixed Pixels, GIScience and Remote Sensing, 45(3): 330–342.
  15. Lu, D., Weng, Q., 2007, A Survey of Image Classification Methods and Techniques for Improving Classification Performance. International Journal of Remote Sensing 28(5): 823-870.
  16. Mather, P.M., 1999, Computer Processing of Remote-Sensed Images, 2. Baskı,
  17. John Wiley and Sons, New York. Melgani, F., Bruzzone, L., 2004, Classification of Hyperspectral Remote Sensing Images w