International Journal of Information Technology & Computer Science ( IJITCS )
‘Since the traditional Support Vector Machine (SVM) algorithm has high classification accuracy at the expense of huge training samples, an automatic hyperspectral image classification method based on subspace partition and SVM (ASP-SVM) algorithm is proposed. In this proposed method, the endmembers were extracted as the representative samples for each class, and a coarse classification result is obtained based on minimum distance clustering. General Sphere Criterion is introduced and applied to the coarse result, and the testing samples is divided into identified samples and unidentified samples. Then, the subspace partition is accomplished according to the probable mixing. Samples which have the highest category confidence in the subspace are selected as the training samples to subdivide the subspace of the unidentified samples to get the final classification. Classification experiment of hyperspectral data illustrates that the proposed approach is satisfied.
: hyperspectral image classification; subspace partition; support vector machines; general sphere criterion
 I. Dopido, M. Zortea, A. Villa, et al., “Unmixing Prior to Supervised Classification of Remotely Sensed Hyperspectral Images”, IEEE Geoscience and Remote Sensing Letters, 2011, 8(4): 760~764.
 G. Moser, S. B. Serpico, “A Markovian generalization of support vector machines for contextual supervised classification of hyperspectral images”, 2010 2nd Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS), 2010, pp: 1~4.
 G. Bilgin, S. Erturk, T. Yildirim, “Unsupervised Classification of Hyperspectral Image Data Using Fuzzy Approaches That Spatially Exploit Membership Relations”, IEEE Geoscience and Remote Sensing Letters, 2008, 5(4): 673~677.
 H. Z. Jiao, Y. F. Zhong, L. P. Zhang, P. X. Li, “Unsupervised Remote Sensing ImageClassification Using an Artificial DNA Computing Artificial DNA Computing”, 2011 Seventh International Conference on Natural Computation (ICNC), July, 2011, 1341~1345.
 W. Z. Liao, A. Pižurica, P. Scheunders, W. Philips, Y. G. Pi, “Semisupervised Local Discriminant Analysis for Feature Extraction in Hyperspectral Images”, IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, JAN., 2013, Vol. 51(1): 184~198.
 X. P. Wen, X. F. Yang, “An Unsupervised Classification Method for Hyperspectral Image Using Spectra Clustering”, IEEE International Symposium on Knowledge Acquisition and Modeling Workshop, 2008, 1117 ~1120.
 C. Cortes, V. Vapnik, “Support Vector Networks”, Machine Learning, 1995,20: 273~297.
 B. Scholkopf, K. Sung, C. Burges, et al., “Comparing Support Vector Machines with Gaussian Kernels to Radial Basis Function Classifiers”, IEEE Transactions on Signal Processing, 1997, 45(11): 2758 ~2765.
 C. J. C. Burges, “Atutorial on Support Vector Machines for Pattern Recognition”, Data Mining and Knowledge Discovery, 1998, 2(2): 955~974.
 K. Tan, P. J. Du, “Hyperspectal remote sensing image classification based on support vector machine”, J. InfraredMillim. Waves, 2008, 27(2): 123~128.
 Y. N. Zhang, L. C. Jiao, “A local adaptive wavelet and gauss neural network synthesis classification system”, Journal of Electronics, 1999,21(3):236~331.
 S. S. Li, Q. J.Tian, “An endmember progressive extraction algorithm for hyperspectral remote sensing image”, Journal o f Remote Sensing , 2009，13(2)：269~275.
 M. Shoshany, F. Kizel, N. S. Netanyahu, et al., “An Iterative Search in End-Member Fraction Space for Spectral Unmixing”, IEEE Geoscience and Remote Sensing Letters, 2011, 8(4): 706~709.
 X. R. Geng, “Target Detection and Classification for Hyperspectral Imagery”, Institute of remote sensing applications, Chinese academy of