OILSEED RAPE PLANTING AREA EXTRACTION BY SUPPORT VECTOR MACHINE USING LANDSAT TM DATA
One parametric classify (Maximum likelihood classify, MLC for short) and two non-parametric classifiers (Adaptive resonance theory mappings and Support vector machines, ARTMAP and SVM for short) were presented in this study. Base on the confusion matrix and the pixels fuzzy analysis, the nonparametric classifier may be a more preferable approach than the parametric classifier for some remote sensing applications and deserves further investigation. The ARTMAP classify represent much best than the rest of classify, especially for the grade of pure pixel (90-100% pureness), Kappa coefficients and overall accuracy were nearly 100%. The higher pureness the pixels were, the better classification accuracy was got.
KeywordsSupport Vector Machine Classification Accuracy Kappa Coefficient Confusion Matrix Parametric Classifier
- C. C. Chang, C. J. Lin. LIBSVM : a library for support vector machines, 2001. Software available at http://www.csie.ntu.edu.tw/∼cjlin/libsvm.
- F. J. Cortijo. A Comparative Study of Classification Methods for Multispectral Images. PhD thesis, DECSAI, Universidad de Granada, Spain. 1995. Available at http://decsai.ugr.es/∼cb/entrada_tesis.html, electronic edition.
- M. J. Barnsley, S. L. Barr. Inferring urban land use from satellite sensor images using kernel-based spatial reclassi. Cation, Photogrammetric Engineering and Remote Sensing, 1996,62: 949–958.Google Scholar
- N. Cristianini, J. Shawe-Taylor. An Introduction to Support Vector Machines. Cambridge University Press, Cambridge, UK, 2000.Google Scholar