高光谱遥感影像SVM分类中训练样本选择的研究

he key of SVM classification is locating an optimal separating hyper-plane and maximizing the margin between the two classes. It is obvious that mixed pixels are much closer to the boundary of classes than pure pixels, much easier to locate the optimal separating hyper-plane. In this article, the reliability and the effect of using mixed pixels as training samples are carried out for hyperspectral image classification by a SVM. Experimental results have shown that hyperspectral remote sensing image classification by a SVM using mixed pixels is feasible, and its accuracy is similar to the accuracy derived from the use of a conventional pure training set, the characteristic of the SVM classification has been demonstrated further that it has a low dependence on the spatial distribution of training samples.