摘要
针对全极化SAR图像在监督分类中存在的人工标注样本费时费力以及多种极化特征未能综合利用等问题,提出一种基于协同训练与集成学习的极化SAR图像半监督分类方法。该方法以支持向量机作为半监督学习的基分类器,通过协同学习机制将多种极化目标分解下的特征有效结合,实现同时利用无标注和有标注样本,最后通过集成学习进一步提高分类模型的泛化能力。在AIRSAR和EMISAR影像上的实验表明,该方法能充分利用不同特征的特点,在较少人工标注的样本下也能获得较高的分类精度。
Supervised classification methods usually require adequate labeled samples which are difficult and time-consuming to obtain,especially in synthetic aperture radar images.In order to make full use of the information of the features by multiple polarization target decomposition methods and the unlabeled samples,A semi-supervised classification method for fully polarimetric SAR imagery based on co-training and ensemble learning is proposed in this paper.The proposed semi-supervised classification method is based on Support Vector Machines（SVM）approach.The features using multiple polarization target decomposition methods are combined by using Co-training mechanism.Both the label and unlabeled samples are utilized for classifying via several co-trained semi-supervised learning models.Finally,the generalization ability of models can be improved by ensemble learning.Experimental results compared to conventional methods on three PolSAR data sets demonstrate that the proposed method can achieve higher classification accuracy with a small number of labeled samples.
出处
《遥感技术与应用》
CSCD
北大核心
2017年第2期380-385,共6页
Remote Sensing Technology and Application
基金
国家自然科学基金项目(41301449)
江苏省测绘地理信息科研项目(JSCHKY201501)
地理空间信息工程国家测绘地理信息局重点实验室经费资助项目(201324)
关键词
极化合成孔径雷达
目标分解
图像分类
半监督学习
协同训练
集成学习
Polarimetric synthetic aperture radar
Target decomposition
Image classification
Semi-supervised learning
Co-training
Ensemble learning
