Classifier Ensembles: Combination of Multiple Classifiers to Improve Classifcation Accuracy of Hyperspectral...

Classifier Ensembles: Combination of Multiple Classifiers to Improve Classifcation Accuracy of Hyperspectral Images

SHARE

Yasser Maghsoudi
Geomatics Engineering Faculty
K.N.Toosi university of Technology
Tehran, Iran
Email: [email protected]

Mohammed Javad Valadan Zoej
Assistant Professor, Department of Geomatic Engineering,
K.N.Toosi University of Technology

Sayyed Bagher Fatemi
Geomatics Engineering Faculty, K.N.Toosi University of Technology
Email: [email protected]

Abstract
Hyperspectral remote sensing with its huge amount of spectral data has greatly extended the scope of traditional remote sensing, it not only allows scientists in environmental and geoscience research communities to obtain much more information about different materials on earth, but also provides many challenges for data analysis tasks. Although this increased spectral resolution leads to material mapping and better environmental investigations, but it is very costly to provide enough training samples for supervised classification. On the other hand, Hughes phenomenon is inevitable with increasing the number of bands.Such factors can directly influence the classification accuracy. In this paper we use classifier ensembles to improve classification accuracy of hyperspectral images. Tow common methods for classifier combinations are:

  • Classifier ensembles with different feature sub sets
  • Classifier ensembles manipulating training samples.

In order to improve the classification performance we integrate these two methods, different feature subsets are taken according to the features correlations, then they are passed on to the K Nearest Neighbor classifiers- with different training samples. Obtained results show that classification accuracy has been improved comparing to the individual classification results.