基于目标分解的面向对象决策树PolSAR影像分类方法研究

发布时间：2018-04-05 18:13

本文选题：PolSAR　切入点：面向对象　出处：《辽宁工程技术大学》2013年硕士论文

【摘要】：全极化合成孔径雷达(Polarization Synthetic Aperture Radar，PolSAR)数据中含有的信息远远多于单极化SAR数据，，但现有的分类方法仅部分地利用了PolSAR数据中的信息，为了将蕴含在PolSAR数据中的信息更加充分地用在PolSAR数据分类中，本文结合目标分解理论、面向对象思想和C5.0决策树算法，提出了一种新的PolSAR数据分类方法。即首先对全极化影像按照不同的目标分解方式进行目标分解，提取包含地物散射机理信息的各种极化参数；然后将提取的极化参数进行分割，将影像的处理单元由像素变为对象；最后使用C5.0决策树算法建立决策规则，实现地物分类。本文采用RADARSAT-2全极化数据进行实验，将本文提出的方法与经典的H/α非监督分类法、监督分类方法和eCognition分类方法等进行了比较，本文提出的方法分类精度达到了83.20%，分别高于其他三种分类方法8.59%、7.67%和1.29%。实验结果表明本文提出的方法具有较高的精度，适用性更强。该方法的提出可以为土地监测、国土资源调查和基础地理信息更新等应用做出贡献。
[Abstract]:The information contained in fully polarized Synthetic Aperture Radarr data is much more than that in single-polarization SAR data, but the existing classification methods only partially utilize the information from PolSAR data.In order to use the information contained in PolSAR data more fully in the classification of PolSAR data, a new method of PolSAR data classification is proposed in this paper, which combines the theory of object decomposition, object-oriented idea and C5.0 decision tree algorithm.First of all, we decompose the whole polarimetric image according to different target decomposition methods, and extract all kinds of polarization parameters which contain the information of scattering mechanism of ground objects, and then segment the extracted polarization parameters.Finally, C5.0 decision tree algorithm is used to establish decision rules to realize the classification of ground objects.In this paper, RADARSAT-2 full polarization data are used to carry out experiments, and the proposed method is compared with classical H / 伪 unsupervised classification, supervised classification and eCognition classification.The classification accuracy of the proposed method is 83.20, which is higher than the other three classification methods 8.597.67% and 1.29% respectively.The experimental results show that the proposed method has higher accuracy and better applicability.This method can contribute to the application of land monitoring, land resources survey and basic geographic information updating.
【学位授予单位】：辽宁工程技术大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P237;P225.1

【参考文献】