Hyperspectral image remote sensing classification using RotBoost

Abstract

Hyperspectral images are able to provide more information because this kind of images have hundreds of spectral bands. In machine learning, the classification of hyperspectral data has several challenges, including high number of dimensions, number of output classes, and limited data references. The solution given to overcome the challenge is to use Ensemble Learning. The benefit of using Ensemble Learning is that we can improve the classification performance of hyperspectral data. One of the Ensemble Learning methods is RotBoost, which is a combination of the Rotation Forest and Adaboost methods. To find out the performance of the RotBoost method, this research used hyperspectral data of vegetation area in Indian pines, Indiana, USA that is provided by NASA’s airborne visible infra-red imaging spectrometer (AVIRIS). The RotBoost method then compared with Rotation Forest method to find a better performance. Confusion matrix used to evaluate the accuracy of each method in the classification. The performance measured is overall accuracy obtained by doing 5-fold cross validation. This experiment also conducted to find the most optimal S (number of base classifiers) and T (number of iteration in Adaboost) values. Experimental results showed that RotBoost produces better accuracy than the Rotation Forest. S and T parameter values are also not very influential on RotBoost accuracy.