基于支持向量机(SVM)单目标SAR图像分割

将考虑SAR图像局部灰度均值和方差及像素空间分布特征等统计量,以灰度共生矩阵产生的纹理统计量为特征向量,经过归一化后进行特征选择,然后输入到支持向量机中进行训练建模,利用支持向量机分类方法,以实现分割结果.最后,将此方法分类结果与传统方法进行了比较,从对比结果可以看出此方法行之有效.

Single Objective Segmentation of SAR Images Based on Support Vector Machine(SVM)

In this paper,some SAR image Statistics indice were considered such as the local gray mean,variance and the spatial distribution characteristics of pixel.Then,the texture statistics from GLCM was taken as features and was normalized,for which feature selection was carried on.In the end,the support vector machine training model(SVM) was constructed,the segmentation was achieved by SVM.This method of classification were compared with the traditional method by test,it showed that it was more effective.