基于目标分解的极化SAR图像SVM监督分类

PolSAR SVM supervised classification method combining with polarimetric target decomposition

鉴于使用单一特征无法获得令人满意的分类效果以及SVM在小训练样本时具有良好的分类性能,提出了基于多种目标分解方法和SVM的极化SAR图像分类方法。首先对原始极化SAR图像使用多种目标分解方法进行处理,得到相应的分量信息,然后在极化SAR图像特征提取的基础上将SVM应用于极化SAR图像分类。通过选取不同的特征信息作为支持向量机的输入,比较其对分类性能的影响,得到最优的用于分类的特征信息组合,其中将相干分解和非相干分解的信息同时用做分类特征能够获得较好的分类效果。利用NASA/JPL实验室AIRSAR系统获取的全极化SAR数据进行实验处理,与Wishart监督分类进行对比,验证了将目标分解信息用做分类特征的有效性,同时与Wishart/H/α和模糊C-均值H/α分类方法进行对比,得到提出的方法具有良好的分类性能。

Because the single feature cannot obtain the satisfactory classification result,and the SVM has the good classification performance with small training sample,this paper proposed a novel method based on the several target decomposition methods and PolSAR classification method based on SVM.First,it decomposed the original PolSAR image into corresponding component information by various target decomposition methods.Then in the basis of feature extraction,it applied the SVM to the PolSAR image classification.By choosing different feature information as the SVM input data and comparing the classification results,the optimal feature information combination could be obtained.Using the NASA/JPL laboratory AIRSAR system data as the experiment data,this paper made a comparison between the proposed method and the Wishart supervised classification to test the performance of the proposed method.The result verifies the proposed method is effective.And then,the paper carried out a further investigation on the comparison among the proposed method,Wishart/H/α method and fuzzy C-means H/α method,it indicates that the proposed method has good classification performance.