基于高分一号遥感影像的绿地信息提取  

Extraction of Urban Green Space Information Based on GF-1 Remote Sensing Images

作　　者：李巍[1] 丁晨旸[2,3] 李萍 

机构地区：[1]黑龙江科技大学矿业工程学院,黑龙江哈尔滨150022 [2]东北林业大学园林学院,黑龙江哈尔滨150001 [3]东北农业大学园艺园林学院,黑龙江哈尔滨150040 [4]山东省东营市园林局,山东东营257100

出　　处：《安徽农业科学》2017年第14期208-210,共3页Journal of Anhui Agricultural Sciences

基　　金：东营园林局横向项目(K-011);黑龙江科技大学教学研究项目(TY14-10)

摘　　要：对高分一号卫星影像进行大气校正、几何校正、裁剪等,利用Libsvm 4.0在Matlab平台里编程进行交叉验证网格法寻优,最终获得支持向量机分类的最佳惩罚系数为45,不敏感系数为0.31。改进支持向量机分类器绿地分类精度为94.6%,该提取精度能满足高分辨率遥感影像在城市绿地动态监测。The atmospheric correction, geometric correction, cutting were conducted on GF-1 satellite images. The cross validation grid optimization was made in Matalb platform by Libsvm 4.0. The best penalty coefficient of support vector machine classifier was 45, and sensitivity coeffi- cient was 0.31. The results showed that the classification accuracy was 94.6%, and the extraction accuracy can meet the high resolution remote sensing images in dynamic monitoring of urban green space.

关 键 词：遥感 高分一号影像 城市绿地 支持向量机分类器 

分 类 号：S127[农业科学—农业基础科学]