论文部分内容阅读
土壤盐渍化不仅严重制约着干旱区农业可持续发展,并且对绿洲生态环境构成了重大威胁。该文提出将多种极化目标分解方法、绕封模型(wrapper feature selector,WFS)特征子集选择方法和支持向量机(support vector machine,SVM)结合起来(简称WFS-SVM),利用全极化合成孔径雷达(polarimetric synthetic aperture radar,Pol SAR)数据实现对土壤盐渍化监测。以新疆干旱区典型绿洲——渭干河-库车河三角洲绿洲(渭-库绿洲)为研究区,对研究区四极化Radarsat-2数据进行多种极化目标分解处理,得到相应的特征参数和特征分量。采用WFS方法进行SVM最佳特征子集的选择,并选出最佳适应度的子集对SVM进行训练。从而构建基于最佳特征子集和最优分类参数的WFS-SVM分类模型,对研究区进行不同程度盐渍地信息(包括重度盐渍地和中-轻度盐渍地)的提取,并结合野外实地考察验证数据,将分类结果与经典的Wishart监督分类方法和一般SVM分类方法进行了对比和验证。结果表明,该方法较大程度地提高了全极化Pol SAR影像在干旱区盐渍地信息提取的精度,相比Wishart监督分类,该方法分类总精度和Kappa系数分别提高了14.12个百分点和0.18,证明了该文所提出的监测方法具有有效性和研究潜力。该成果也将促进Pol SAR数据在干旱区盐渍化监测中发挥更大的作用。
Soil salinization not only severely restricts the sustainable development of agriculture in arid areas, but also poses a serious threat to the ecological environment of oases. In this paper, a combination of multiple polarization target decomposition methods, WFS feature subset selection method and support vector machine (SVM) are proposed, Data of Synthetic Aperture Radar (SAR) with Pol SAR Data to Monitor Soil Salinization. Taking the typical oasis in the arid region of Xinjiang-Weigan-Kuqa River delta oasis (Wei-Lv Oasis) as the research area, we carried out a variety of polarization target decomposition of the quadrupole Radarsat-2 data in the study area and obtained the corresponding characteristics Parameters and feature components. The WFS method is used to select the best feature subset of SVM and a subset of best fitness is selected to train SVM. Therefore, a WFS-SVM classification model based on the best subset of features and the optimal classification parameters is constructed to extract salinity information (including heavily salted land and medium-light salted land) in the study area in combination with Field field validation data, the classification results and the classic Wishart supervised classification methods and general SVM classification methods were compared and verified. The results show that the proposed method can greatly improve the accuracy of extracting information from salty ground of the polly polarized Pol SAR image in the arid region. Compared with Wishart classification, the proposed method improves the classification accuracy and Kappa coefficient by 14.12 percentage points and 0.18 , Which proves the validity and research potential of the monitoring method proposed in this paper. This result will also promote Pol SAR data to play a greater role in the monitoring of salinization in arid zones.