Hyperspectral image classification based on Boosted Rotation Forest

摘要

Recent literature has revealed that Rotation Forest (RoF), a novel classifier ensemble approach, results in excellentclassification performance for hyperspectral remote sensing images. To further promote the performance of RoF, we combineBoosting and RoF to propose a Boosted Rotation Forest (BRoF) classifier. Using classification and regression tree (CART) as thebase classifier, we investigated the performance of BRoF on Airborne Visible/Infrared Imaging Spectrometer (AVIRIS)hyperspectral image and compared it with Boosting and RoF. Experimental results demonstrate the better performance of BRoFthan Boosting and RoF.