基于多种植被指数进行春小麦提取  

Spring Wheat Extraction Based on Multiple Vegetation Indices

作　　者：邹健 刘沼辉 刘晓 ZOU Jian;LIU Zhaohui;LIU Xiao(College of Geomatics,Shandong University of Science and Technology,Qingdao Shandong 211590,China)

机构地区：[1]山东科技大学测绘科学与工程学院

出　　处：《北京测绘》2019年第10期1210-1213,共4页Beijing Surveying and Mapping

摘　　要：文章以湟中县为研究区域,采用高分一号影像数据。在全生育期波谱特征曲线分析基础上,提取作物的NDVI曲线特征,并结合春小麦全生育周期内RVI(比值植被指数)与NDGI(归一化差异绿度植被指数)特征变化设计了春小麦提取模型。最后通过精度自检得出基于NDVI时序影像其总体精度达到93.81%,kappa系数为0.875,基于3种植被指数时序影像提取总体精度达到95.45%,kappa系数达到0.922。从分类精度可以看出,利用中高分辨率遥感卫星影像,在作物NDVI时间序列变化规律分析的基础上,通过其他植被指数和辅助数据可以精确的进行大面积农作物的分类与提取。This study takes Huangzhong County of Qinghai province as the study area and use GF-1 image data.On the basis of the analysis of the spectrum characteristics of the whole growth period,the characteristics of the NDVI curve of the crop type are extracted.The spring wheat extraction model was designed by analyzing the characteristics of RVI(ratio vegetation index)and NDGI(normalized difference green vegetation index)in the whole reproductive cycle of spring wheat.Finally,it is concluded that the overall accuracy of the NDVI sequential image is 93.81%and the kappa coefficient is 0.875.Based on three kinds of vegetation index timing images,the overall accuracy reaches 95.45%,and the kappa coefficient reaches 0.922.It can be seen from the classification accuracy that the classification and extraction of large-area crops can be accurately carried out by using other vegetation index sequences and auxiliary data on the basis of the analysis of the change law of the time series of the crop NDVI by using the medium-high resolution remote sensing satellite images.

关 键 词：农作物提取 多植被指数 时间序列影像 决策树分类 混淆矩阵 湟中县 

分 类 号：P232[天文地球—摄影测量与遥感]