基于支持向量机的遥感图像分类方法

TM Image Classification Based on Support Vector Machine

为了提高遥感图像分类的精度,弥补传统最大似然分类方法所固有的分类时样本不足的缺陷,提出了一种基于支持向量机、光谱特征和纹理特征相结合的遥感图像分类方法。采用ETM数据,按照其所提方法进行了具体分类实验,并将实验结果与最大似然法分类的结果进行了比较分析。结果表明,利用基于支持向量机的方法进行遥感图像分类,精度明显优于最大似然法分类的精度。利用光谱特征与纹理特征相结合进行分类比单纯运用光谱特征进行分类效果要好。

In order to improve the accuracy of remote sensing image classification and compensate the weakness of maximum likelihood classifier, this paper puts forward a new classification method, which is based on Support Vector Machine （SVM）. This method combines the spectrum features with texture ones. According to the method classification test is done with ETM data, and the accuracy is compared with the one of maximum likelihood classifier. The results indicate that the accuracy obtained from the new method is better than the other＇ s, and combining spectrum feature and texture one is better than the one of only using one kind of feature.