The airborne hyperspectral image classification based on the random forest algorithm
Authors:
- 3 November 2016
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 2280-2283
Abstract
The hyperspectral image has the advantages of wide spectral range and the high spectral resolution, and is widely applied in the terrain classification. In this paper, we study the airborne hyperspectral image classification methods using the airborne hyperspectral image. Considering the hyperspectral image has amounts of bands and there is redundancy among the bands, the principle component analysis is adopted to delete the redundant bands and the noise bands. And then the random forest decision tree algorithm is employed to classify the image. Through calculating the importance coefficient, the contribution of each band to the image classification is acquired. Finally, we design and develop the classification program using the C++ language with the OpenCV library to classify the hyperspectral image of Zhangye city collected by airborne AISA EAGLET sensor. The result of experiment indicates the validity and the robustness of random forest algorithm.Keywords
This publication has 3 references indexed in Scilit:
- Classification of hyperspectral remote-sensing data with primal SVM for small-sized training dataset problemAdvances In Space Research, 2008
- Contaminant Classification of Poultry Hyperspectral Imagery using a Spectral Angle Mapper AlgorithmBiosystems Engineering, 2007
- Random ForestsMachine Learning, 2001