摘要
针对Landsat-8 OLI和GF-1 WFV传感器参数的特点,选择支持向量机(SVM)分类方法分别对咸宁市同一时段的Landsat-8遥感影像和GF-1遥感影像进行土地利用分类研究。结果表明,Landsat-8在耕地与林地、水域与裸地可分离性方面高于GF-1,提取的林地面积占比和耕地面积占比更接近于真实值;Landsat-8和GF-1的分类总精度分别为85.76%和88.38%,Kappa系数分别为0.807 1和0.820 4,说明GF-1的分类效果好于Landsat-8;GF-1具有较高的分辨率优势,对分布零散的地物识别效果优于Landsat-8。
According to the characteristics of Landsat-8 OLI and GF-1 WFV sensor parameters, the support vector machine （SVM） classification method was used to classify Landsat-8 remote sensing images and GF-1 remote sensing images at the same time in Xianning City. The results showed that the separation of water area Landsat-8 in the cultivated land,forest land, and bare land was higher than that of GF-1, and the propor-tion of extracted forest land and cultivated land was closer to the real value. The classification total accuracy of Landsat-8 and GF-1 were 85. 76% and 88. 38% respectively, and Kappa coefficients were 0. 807 1 and 0. 820 4 respectively. The classification effect of GF-1 was better than that of Landsat-8. GF-1 had higher resolution advantages, and the classification effect of the fragmented landform type was better than that of Landsat-8.
出处
《安徽农业科学》
CAS
2017年第31期213-215,237,共4页
Journal of Anhui Agricultural Sciences
基金
湖北省教育厅人文社会科学研究青年项目(15Q217)
湖北科技学院校级科研项目(2016-18X058)
关键词
遥感影像
监督分类
可分离度
Kappa系数
Remote sensing images
Supervised classification
Separability
Kappa coefficient
