基于时变特征的多时相PolSAR农作物分类方法
发布时间：2018-09-04 14:32
【摘要】：获取农作物分类信息是极化合成孔径雷达(Polarimetric synthetic aperture radar,PolSAR)的重要应用之一,然而单时相PolSAR数据能够提供的信息十分有限,而且单时相数据的获取时间也会影响农作物分类精度。随着技术的发展,出现了大量的机载和星载PolSAR系统,这些系统能够获取目标重复观测的PolSAR数据,这为多时相PolSAR的数据分析和应用提供了可能。本文以多时相PolSAR农作物分类为出发点,通过利用不同农作物的极化散射特性的变化特性来提高分类精度。首先,基于极化散射特性分解原理分析了不同农作物在生长过程不同时期所呈现的散射特性变化规律,在此基础上定义了一个新的参数描述其散射特性的变化特性。其次,基于这一新参数提出了一种多时相PolSAR农作物监督分类算法。最后,通过对欧洲空间局所提供的基于Radarsat-2实测仿真生成的Sentinel-1数据处理结果表明,相比于基于复Wishart分布的监督分类算法,农作物的整体分类精度提高了约4个百分点,当农作物种类合并为4类时,整体分类精度提高了约6个百分点。
[Abstract]:Obtaining crop classification information is one of the important applications of Polarimetric synthetic Aperture Radar (Polarimetric synthetic aperture radar,PolSAR). However, the single phase PolSAR data can provide very limited information, and the acquisition time of the single time phase data will also affect the precision of crop classification. With the development of technology, a large number of airborne and spaceborne PolSAR systems have emerged. These systems can obtain the PolSAR data of target repeated observation, which provides the possibility for the data analysis and application of multitemporal PolSAR. In this paper, the classification accuracy is improved by using the polarization scattering characteristics of different crops as the starting point of multi-phase PolSAR crop classification. Firstly, based on the decomposition principle of polarimetric scattering characteristics, the variation of scattering characteristics of different crops in different periods is analyzed, and a new parameter is defined to describe the variation characteristics of scattering characteristics. Secondly, based on this new parameter, a multi-temporal PolSAR crop classification algorithm is proposed. Finally, the processing results of Sentinel-1 data based on Radarsat-2 simulation provided by the European Space Agency show that compared with the supervised classification algorithm based on complex Wishart distribution, the overall classification accuracy of crops is improved by about 4 percentage points. When the crop species are merged into 4 categories, the overall classification accuracy is improved by about 6 percentage points.
【作者单位】： 西北农林科技大学机械与电子工程学院;农业部农业物联网重点实验室;联邦科学与工业研究组织数据处理研究所;
【基金】：国家自然科学基金项目(41301450、61701416) 卫星测绘技术与应用国家测绘地理信息局重点实验室开放基金项目(KLSMTA-201501)
【分类号】：S127
本文编号：2222440
[Abstract]:Obtaining crop classification information is one of the important applications of Polarimetric synthetic Aperture Radar (Polarimetric synthetic aperture radar,PolSAR). However, the single phase PolSAR data can provide very limited information, and the acquisition time of the single time phase data will also affect the precision of crop classification. With the development of technology, a large number of airborne and spaceborne PolSAR systems have emerged. These systems can obtain the PolSAR data of target repeated observation, which provides the possibility for the data analysis and application of multitemporal PolSAR. In this paper, the classification accuracy is improved by using the polarization scattering characteristics of different crops as the starting point of multi-phase PolSAR crop classification. Firstly, based on the decomposition principle of polarimetric scattering characteristics, the variation of scattering characteristics of different crops in different periods is analyzed, and a new parameter is defined to describe the variation characteristics of scattering characteristics. Secondly, based on this new parameter, a multi-temporal PolSAR crop classification algorithm is proposed. Finally, the processing results of Sentinel-1 data based on Radarsat-2 simulation provided by the European Space Agency show that compared with the supervised classification algorithm based on complex Wishart distribution, the overall classification accuracy of crops is improved by about 4 percentage points. When the crop species are merged into 4 categories, the overall classification accuracy is improved by about 6 percentage points.
【作者单位】： 西北农林科技大学机械与电子工程学院;农业部农业物联网重点实验室;联邦科学与工业研究组织数据处理研究所;
【基金】：国家自然科学基金项目(41301450、61701416) 卫星测绘技术与应用国家测绘地理信息局重点实验室开放基金项目(KLSMTA-201501)
【分类号】：S127
【相似文献】
相关期刊论文 前4条
1 牛朝阳;马德宝;张向锋;;基于H/α/A分解的PolSAR农作物分类方法研究[J];信息工程大学学报;2006年04期
2 彭光雄;宫阿都;崔伟宏;明涛;陈锋锐;;多时相影像的典型区农作物识别分类方法对比研究[J];地球信息科学学报;2009年02期
3 袁竞;;宁波市2000年多时相植被遥感[J];安徽农业科学;2011年20期
4 张国庆;黄楠;宋茜;莫红;;利用多源多时相卫星影像对黑龙江省耕地水涝灾害的监测[J];黑龙江农业科学;2014年06期
，本文编号：2222440
本文链接：https://www.wllwen.com/kejilunwen/nykj/2222440.html
