中文摘要：

为了验证噪声对支持向量机分类器性能的影响,对＂SVM可以有效用于含噪声和不确定性数据＂这一观点进行定量分析评价,采用国产OMISII传感器获得的高光谱遥感数据进行了试验,为了更好地比较SVM分类器的抗噪性,先对原始数据进行支持向量机分类,然后在高光谱遥感影像中人为添加不同比例的椒盐噪声和条带噪声,然后进行支持向量机分类,并与传统的光谱角制图和最小距离分类算法进行比较。结果表明支持向量机具有明显的抗噪性,其分类性能受噪声影响较小,是一种有效的高光谱遥感影像分类器。

英文摘要：

The performance of SVM for hyperspectral image classification has been examined from a wide range of perspectives. This paper aims to evaluate the responses of SVM classifier to noises or uncertainty in original data, or the impacts of noise on SVM classifier, The research was undertaken using hyperspectral image captured by OMISII sensor with 64 bands. Firstly, the original data were classified by SVM. Two types of noises, salt-pepper noise and striping, were added into the original data and then the mixed data were classified by SVM. The classification results of SVM, compared with traditional classifiers including spectral angle mapper （SAM） and minimum distance classifier, indicated that SVM classifier are more effective to alleviate the effects of noises.