| 面向对象的旱区植被遥感精细分类研究 |
| |
| 引用本文： | 张文博,孔金玲,杨园园,李彤.面向对象的旱区植被遥感精细分类研究[J].测绘科学,2021,46(1):136-140,183. |
| |
| 作者姓名： | 张文博 孔金玲 杨园园 李彤 |
| |
| 作者单位： | 长安大学地球科学与资源学院,西安710054;长安大学地质工程与测绘学院,西安710054;长安大学地球科学与资源学院,西安710054;长安大学地球科学与资源学院,西安710054 |
| |
| 基金项目： | 新疆乌伦古河流域水文地质环境地质调查项目 |
| |
| 摘 要： | 针对旱区植被分类尺度过大、种群无法准确提取的问题,该文提出了面向对象的CFS-RF分类模型,即利用CFS算法对先验样本数据集进行特征优选,结合随机森林构建分类规则,完成分类过程。以新疆阿勒泰为研究区,利用GF-2数据,通过CFS、ReliefF两种不同特征选择方法和J48、SVM、RF 3种分类算法构造出6种面向对象分类方案来实现小尺度植被种群提取。结果表明,经过特征选择,上述分类方案的精度和效率均得到了提升。其中,CFS-RF算法最优,总体精度达到92.41%,Kappa系数为0.90,更适用于旱区植被遥感精细分类。
|
| 关 键 词： | 高分影像 旱区植被 特征优选 面向对象分类 |
Fine vegetation classification of remote sensing in arid areas based on object-oriented method |
| |
| ZHANG Wenbo,KONG Jinling,YANG Yuanyuan,LI Tong.Fine vegetation classification of remote sensing in arid areas based on object-oriented method[J].Science of Surveying and Mapping,2021,46(1):136-140,183. |
| |
| Authors: | ZHANG Wenbo KONG Jinling YANG Yuanyuan LI Tong |
| |
| Institution: | (School of Earth Science and Resources,Chang’an University,Xi’an 710054,China;School of Geological Engineering and Geomatics,Chang’an University,Xi’an 710054,China) |
| |
| Abstract: | Aiming at the problem that the classification scale of vegetation in arid areas was too large and the species could not be accurately extracted,this paper proposed an object-oriented algorithm named CFS-RF,CFS algorithm was used to optimize the features of prior sample data sets,Combining with random forest,the classification rules were constructed and the classification process was completed.In this paper,taking Altay,Xinjiang province as the research area,using GF-2data,six object-oriented classification schemes were constructed by combining two different feature selection methods,CFS and ReliefF,with three classification algorithms,J48,SVM and RF,to achieve the extraction of small-scale vegetation species.The result showed that through feature selection,the accuracy and efficiency of the above classification scheme had been improved,and the CFS-RF algorithm was the best,the overall accuracy reached 92.41%,and the Kappa coefficient was 0.90,it was more suitable for fine vegetation classification using remote sensing in arid areas. |
| |
| Keywords: | high-resolution image vegetation in arid areas feature selection object oriented classification |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
