|国家预印本平台
首页|基于协同训练的极化SAR目标分类

基于协同训练的极化SAR目标分类

Polarimetric SAR Classification based on Co-Training

中文摘要英文摘要

地物分类是极化SAR(POLSAR)数据应用中的一个重要问题。如何有效地利用极化特征进行地物分类是目前极化SAR领域研究的热点问题。半监督学习(Semi-Supervised Learning,SSL)能够同时利用大量未标记样本和少量有标记样本信息。协同训练算法(Co-Training)作为半监督学习的重要算法之一,可以更好的利用多视角特征。本文对基于协同训练的极化SAR地物分类问题展开了研究,将协同训练与支撑矢量机(SVM)方法结合,提出了基于SVM的协同训练策略,并将其应用于极化SAR图像地物分类。实验结果表明,该方法相比传统协同训练算法具有较好的分类精度。

errain classification plays an important role in POLSAR image data application. It is a hot topic that how to classify the image data through using polarimetric features properly. SSL can use both major unlabeled samples and minor labeled ones. Co-training, one of the most important algorithms of SSL, utilizes the multi-view information. In this paper, the focus of the study is the terrain classification of POLSAR based on co-training. This paper proposes the SVM based co-training strategy by combining co-training and SVM together, which are then applied to terrain classification of POLSAR image. The result shows that its accuracy is much better than other traditional algorithms’.

张青、王爽、焦李成

遥感技术

图像处理半监督学习协同训练极化SAR图像分类

Image ProcessingSemi-Supervised LearningCo-TrainingPolarimetric SARImage Classification

张青,王爽,焦李成.基于协同训练的极化SAR目标分类[EB/OL].(2012-01-17)[2025-08-23].http://www.paper.edu.cn/releasepaper/content/201201-616.点此复制

评论