测绘学报

• •    

基于SVM和PWC的遥感影像混合像元分解

李慧,王云鹏,李岩,王兴芳   

  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2011-12-28 发布日期:2019-01-01

Mixed Pixels Classification of Remote Sensing Images Based on Support Vector Machines and Pairwise Coupling

Hui li, Yunpeng Wang, Yan li, Xingfang Wang   

  • Received:1900-01-01 Revised:1900-01-01 Online:2011-12-28 Published:2019-01-01

摘要: 支持向量机(Support Vector Machines, SVM)与两两配对(Pairwise Coupling, PWC)方法结合可分解遥感影像混合像元。首先支持向量机的输出值转化为两两配对的后验概率,再由两两配对的概率值求得多类后验概率,最终像元所属类别的后验概率作为地物的组分信息。利用多波段遥感数据验证了此方法的可行性,并将结果与线性分解模型进行比较。结果表明,SVM与PWC结合进行混合像元分解在准确性方面,明显优于一般线性模型的精度,并且此方法用于图像分类中也可以有好的结果。

Abstract: Support Vector Machines (SVM) with pairwise coupling (PWC) method is designed to estimate abundance fractions of materials in remote sensing image. PWC method maps the SVM outputs into posteriori probabilities which are regarded as abundances of each material. Multi-channel remote sensing images data are used to validate the method. The experiment result shows that the method can provide better result of abundance estimation as compared with general linear-model and it can get good result in image classification.