254
10.S
Summary
10: Pakorn Watanachaturaporn, Manoj K. Arora
In this chapter, we have considered the application of SVMs for the classification
of multi and hyperspectral remote sensing datasets. We also investigated the
effect of various parameters on the accuracy of SVM based classification. It
is clear from the results that with an appropriate selection of the multiclass
method, optimizer and the kernel function, accuracy of the order of 95% can be
obtained for the classification of multi and hyperspectral image classification.
The penalty value (the C value) has an important bearing on the performance
of any SVM classifier. However, this may have to be selected by trial and error
for a given dataset.
References
Byun H, Lee SW (2003) A survey on pattern recognition applications of support vector
machines. International Journal of Pattern Recognition and Artificial Intelligence 17:
459-486
Chang CC, Lin CJ (2002) LIBSVM: a library for support vector machines
(http://www.csie.ntu.edu.tw/~cjlin/libsvm )
Gualtieri JA, Cromp RF (1998) Support vector machines for hyperspectral remote sensing classification. In: Merisko RJ (ed), Proceeding of the SPIE, 27th AIPR Workshop,
Advances in Computer Assisted Recognition, 3584, pp 221-232
Gualtieri JA, Chettri SR, Cromp RF, Johnson LF (1999) Support vector machine classifiers
as applied to AVIRIS data. Proceedings of Summaries of the Eighth JPL Airborne Earth
Science Workshop (ftp:/ /popo.jpl.nasa.gov/pub/docs/workshops/99 _docs/31.pdf)
Huang C, Davis LS, Townshend JR (2002) An assessment of support vector machines for
land cover classification. International Journal of Remote Sensing 23(4): 725-249
Joachims T (2002) The SVM1ig ht package (http://svmlight.joachims.org or ftp://ftp-aLcs.unidortmund.de)
Keerthi SS (2002) Efficient tuning ofSVM hyperparameters using radius/margin bound and
iterative algorithms. IEEE Transactions on Neural Networks 13: 1225-1229
Lee C, Landgrebe DA (1993) Feature Extraction and Classification Algorithms for High
Dimensional Data. Technical Report, TR-EE 93-1, School of Electrical Engineering,
Purdue University
Mangasarian OL, Musicant D (2000) Lagrangian support vector machines, Technical Report
(0006), Data Mining Institute, Computer Science Department, University of Wisconsin,
Madison, Wisconsin (ftp://ftp.cs.wisc.edu/pub/dmi/tech-reportl0006.ps)
Melgani F, Bruzzone L (2002) Support vector machines for classification of hyperspectral
remote-sensing images. International Geoscience and Remote Sensing Symposium,
IGARSS'02, CD.
Shah CA, Watanachaturaporn P, Arora MK, Varshney PK (2003) Some recent results on hyperspectral image classification. Proceedings of IEEE Workshop on Advances in Techniques for Analysis of Remotely Sensed Data, NASA Goddard Space Flight Center,
Greenbelt, MD, CD
Tadjudin S, Landgrebe DA (1998) Classification of High Dimensional Data with Limited
Training Samples. PhD thesis, School of Electrical Engineering and Computer Science,
Purdue University.
10.S
Summary
10: Pakorn Watanachaturaporn, Manoj K. Arora
In this chapter, we have considered the application of SVMs for the classification
of multi and hyperspectral remote sensing datasets. We also investigated the
effect of various parameters on the accuracy of SVM based classification. It
is clear from the results that with an appropriate selection of the multiclass
method, optimizer and the kernel function, accuracy of the order of 95% can be
obtained for the classification of multi and hyperspectral image classification.
The penalty value (the C value) has an important bearing on the performance
of any SVM classifier. However, this may have to be selected by trial and error
for a given dataset.
References
Byun H, Lee SW (2003) A survey on pattern recognition applications of support vector
machines. International Journal of Pattern Recognition and Artificial Intelligence 17:
459-486
Chang CC, Lin CJ (2002) LIBSVM: a library for support vector machines
(http://www.csie.ntu.edu.tw/~cjlin/libsvm )
Gualtieri JA, Cromp RF (1998) Support vector machines for hyperspectral remote sensing classification. In: Merisko RJ (ed), Proceeding of the SPIE, 27th AIPR Workshop,
Advances in Computer Assisted Recognition, 3584, pp 221-232
Gualtieri JA, Chettri SR, Cromp RF, Johnson LF (1999) Support vector machine classifiers
as applied to AVIRIS data. Proceedings of Summaries of the Eighth JPL Airborne Earth
Science Workshop (ftp:/ /popo.jpl.nasa.gov/pub/docs/workshops/99 _docs/31.pdf)
Huang C, Davis LS, Townshend JR (2002) An assessment of support vector machines for
land cover classification. International Journal of Remote Sensing 23(4): 725-249
Joachims T (2002) The SVM1ig ht package (http://svmlight.joachims.org or ftp://ftp-aLcs.unidortmund.de)
Keerthi SS (2002) Efficient tuning ofSVM hyperparameters using radius/margin bound and
iterative algorithms. IEEE Transactions on Neural Networks 13: 1225-1229
Lee C, Landgrebe DA (1993) Feature Extraction and Classification Algorithms for High
Dimensional Data. Technical Report, TR-EE 93-1, School of Electrical Engineering,
Purdue University
Mangasarian OL, Musicant D (2000) Lagrangian support vector machines, Technical Report
(0006), Data Mining Institute, Computer Science Department, University of Wisconsin,
Madison, Wisconsin (ftp://ftp.cs.wisc.edu/pub/dmi/tech-reportl0006.ps)
Melgani F, Bruzzone L (2002) Support vector machines for classification of hyperspectral
remote-sensing images. International Geoscience and Remote Sensing Symposium,
IGARSS'02, CD.
Shah CA, Watanachaturaporn P, Arora MK, Varshney PK (2003) Some recent results on hyperspectral image classification. Proceedings of IEEE Workshop on Advances in Techniques for Analysis of Remotely Sensed Data, NASA Goddard Space Flight Center,
Greenbelt, MD, CD
Tadjudin S, Landgrebe DA (1998) Classification of High Dimensional Data with Limited
Training Samples. PhD thesis, School of Electrical Engineering and Computer Science,
Purdue University.