首页> 外文会议>IEEE International Geoscience and Remote Sensing Symposium >Classifying Multi-Channel Polsar Images Base on Polarization Signature
【24h】

Classifying Multi-Channel Polsar Images Base on Polarization Signature

机译:基于极化特征的多通道Polsar图像分类

获取原文

摘要

The conventional classifying methods use polarimetric information in the restrict number of polarization basis and frequency for polarimetric data classification. At the same time, different polarization and frequencies are sensitive to different surface scales and scattering mechanisms respectively. The combined use of different frequencies on the various polarization basis can improve the classification accuracy. This paper proposes a new classification approach for multifrequency polarimetric SAR (PolSAR) data based on polarization signature. At the first step, polarization signature is generated from coherency matrix. In the second step, the Random Forest (RF) classifier is used for classifying PolSAR data. Then, in order to combine the output of RF for incorporated three frequencies, majority voting method is used. An AIRSAR image from Sault Ste. Marie city in Ontario, Canada was chosen for this study. The results showed that different classes of land cover at different frequencies have a various performance accuracy compared to single-frequency data. Using polarization signature on three frequencies (C, L, and P) can improve classification accuracy near to 4%.
机译:传统的分类方法在限制偏振数和频率的限制次数中使用偏振次数,用于偏振数据分类。同时,不同的极化和频率分别对不同的表面尺度和散射机构敏感。不同频率对各种极化基础的结合使用可以提高分类精度。本文提出了一种基于极化签名的多频偏振SAR(POLSAR)数据的新分类方法。在第一步中,从一致性矩阵生成极化签名。在第二步中,随机林(RF)分类器用于分类POLSAR数据。然后,为了结合RF的输出,用于结合的三个频率,使用大多数表决方法。来自sault ste的航空图像。在加拿大安大略省的玛丽市被选为这项研究。结果表明,与单频数据相比,不同频率的不同频率的陆地覆盖具有各种性能精度。在三个频率(C,L和P)上使用偏振签名可以提高靠近4%的分类精度。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号