一种基于支持向量机和主成分分析的多光谱图像的分类方法

多光谱图像分类方面,由于普通的SVM方法没有考虑多光谱图像具有高维度和冗余的特点,因此难以实现令人满意的分类精度.本文提出了一种基于SVM和主成分分析相结合的多光谱的图像分类方法.并用5幅6波段两类地形的多光谱图像进行实验.实验结果表明,这种分类方法与普通的SVM方法相比提高了多光谱图像的分类精度.

A multispectral image classification method based on the SVM and PCA

In the field of multispectral image classification,the feature of high dimensions and redundancy of the multispectral images is not considered by the common SVM method,so the classification accuracy is not satisfying.This paper proposed a method based on the SVM and PCA to classify the multispectral image.We experimented on 5 multispectral image of 6 bands which contained 2 classes of terrains.The experiment's result showed that this method could improve the segmentation accuracy.