Multi-class support vector machine classifier based on separability measure for hyperspectral remote sensing image classification

According to SVM theory and the separability measure of hyperspectral data, we put forward a novel binary tree multi-class SVM classifier based on separability between different classes, constructed different multi-class SVM classifiers and tested their accuracy by experimented the hyperspectral image with the 64 bands OMISII data and Hyperion hyperspectral data. The experimental results show that the novel binary tree classifier has the highest accuracy than the other multi-class SVM classifiers and some traditional classifiers (spectral angle mapping classification and minimum distance classification). Use of the novel binary tree multi-class SVM classifier based on separability measure is a novel approach which improves the accuracy of hyperspectral image classification and expands the possibilities for scientific interpretation and application.