The Classification of Diseased Trees by Using kNN and MLP Classification Models According to the Satellite Imagery

Abstract

In this study, the Japanese Oak and Pine Wilt in forested areas of Japan was classified into two group as diseased trees and all other land cover area according to the 6 attributes in the spectral data set of the forest. The Wilt Data Set which was obtained from UCI machine learning repository database was used. Weka (Waikato Environment for Knowledge Analysis) software was used for classification of areas in the forests. The classification success rates and error values were calculated and presented for classification data mining algorithms just as Multilayer Perceptron (MLP) and k-Nearest Neighbor (kNN). In MLP neural networks the classification performance for various numbers of neurons in the hidden layer was presented. The highest success rate was obtained as 86.4% when the number of neurons in the hidden layer was 10. The classification performance of kNN method was calculated for various counts of neighborhood. The highest success rate was obtained as 72% when the count of neighborhood number was 2.

Keywords

• k - Nearest Neighbor,Multilayer Perceptron Neural Network; Weka; Classification; Remote Sensing

References

1. Han J., Pei J., Kamber M., Data Mining: Concepts and Techniques, 3. Edition, Morgan Kaufmann, 2012, p.740.
2. Hegland M., Data mining techniques, Acta Numerica, 2001, Vol. 10, pp 313-355.
3. Michael J. Shaw M.J., Subramaniam C., Tan G.W., Welge M.E., Knowledge management and data mining for marketing, Decision Support Systems, 2001, vol. 31(1), pp.127-137.
4. Frank E., Hall M., Trigg L., Holmes G., Witten I. H., Data mining in bioinformatics using Weka, Bioinformatics Applications Note, 2004, Vol.20(15), pp. 2479-2481.
5. Kurvonen L., Hallikainen M. T., Textural Information of Multitemporal ERS-1 and JERS-1 SAR Images with Applications to Land and Forest Type Classification in Boreal Zone, IEEE Transactions On Geoscience And Remote Sensing, 1999, VOL. 37(2), pp. 680-689.
6. Kosaka, N., Akiyama, T., Tsai, B., Kojima, T., Forest type classification using data fusion of multispectral and panchromatic high-resolution satellite imageries, In International Geoscience and Remote Sensing Symposium, 2005, Vol. 4, pp. 2980 2983.
7. Yang X., Rochdi N., Zhang J., Banting J., Rolfson D., King C., Purdy B., Mapping tree species in a boreal forest area using RapidEye and LiDAR data, In Geoscience and Remote Sensing Symposium (IGARSS), 2014 IEEE International, pp. 69-71.
8. Shackleton C.M., Shackleton S.E., Buiten E., Bird N., The importance of dry woodlands and forests in rural livelihoods and poverty alleviation in South Africa, Forest Policy and Economics, 2007, Vol. 9(5), Pages 558–577.

9. Shvidenko A., Barber C.V., Persson R., Forest and woodland systems, Ecosystems and Human Well-being: Current State and Trends, 2005, Vol 1, pp. 587–621.
10. Sturrock R. N., Frankel S. J., Brown A. V., Hennon P. E., Kliejunas J. T., Lewis K. J., Worrall J. J., Woods A. J., Climate change and forest diseases, Plant Pathology, 2011, vol. 60(1), pp. 133-149.
11. Johnson, B., Tateishi, R., Hoan, N., A hybrid pansharpening approach and multiscale object-based image analysis for mapping diseased pine and oak trees, International Journal of Remote Sensing, 2013, vol. 34(20), pp. 6969-6982.
12. Hall M. , Frank E., Holmes G., Pfahringer B., Reutemann P., Witten I. H., The WEKA Data Mining Software: An Update, SIGKDD Explorations, 2009, Vol. 11(1), pp. 10-18.
13. Arora R., Suman S., Comparative Analysis of Classification Algorithms on Different Datasets using WEKA, International Journal of Computer Applications, 2012, Vol. 54(13), pp. 21-25.
14. Wang J., Neskovic P., Cooper L. N., Improving nearest neighbor rule with a simple adaptive distance measure, Pattern Recognition Letters, 2007, Vol. 28(2), pp. 207-213.
15. Zhou Y., Li Y., Xia S., An improved KNN text classification algorithm based on clustering, Journal of computers, 2009, Vol. 4(3), pp. 230-237.