文献类型：期刊文献

中文题名：基于NDVI分割与面向对象的东洞庭湖湿地植被信息提取技术

英文题名：Extraction of Vegetation Information Based on NDVI Segmentation and Object-oriented Method

作者：乔婷[1] 张怀清[1] 陈永富[1] 凌成星[1]

第一作者：乔婷

机构：[1]中国林业科学研究院资源信息研究所

年份：2013

卷号：28

期号：4

起止页码：170-175

中文期刊名：西北林学院学报

外文期刊名：Journal of Northwest Forestry University

收录：CSTPCD;;北大核心:【北大核心2011】；CSCD:【CSCD_E2013_2014】；

基金：林业科学技术推广项目([2012]16号);国家863计划课题(2009AA122003-L);国家重大专项(E0305/1112/02)

语种：中文

中文关键词：SPOT-5影像;面向对象分类;NDVI分割;信息提取

外文关键词：SPOT-5 image; object-oriented methodology; NDVI segmentation; information extractio

分类号：S771.8

摘要：针对东洞庭湖湿地植被的分布现状,为提高湿地植被信息提取的精度和效率,应用高分辨率的SPOT-5影像数据,在完成图像预处理的基础上,将NDVI应用到多尺度分割中,结合基于隶属度函数和阈值的面向对象的分类方法对东洞庭湖湿地植被信息进行提取;与此同时,以相同分类方法对未辅以NDVI分割的图像进行植被提取,并与最大似然监督分类法提取的结果进行对比。结果显示,辅以NDVI分割的面向对象信息提取的总分类精度达到了87.69%,Kappa系数达到0.86;未辅以NDVI的总分类精度为82.69%,Kappa系数为0.80;最大似然监督分类总分类精度71.92%,Kappa系数0.67;其总分类精度分别提高了5.00%、15.77%,Kappa系数分别提高了0.06、0.19。可见,该方法可以有效提高湿地植被的提取精度,为湿地植被资源进一步的监测和保护奠定了基础。
In order to increase the extraction accuracy and efficiency of vegetation information,according to the vegetation distribution in the East Dongting Lake wetland,SPOT-5 remote sensing images were used as the data sources.After the images were pre-processed,NDVI was applied to the multi-scale segmentation to optimize vegetation extraction precise.Based on analyzing spectral,geometric and some other characteristics,the extraction of the East Dongting Lake wetland vegetation information was conducted by using membership functions and threshold of object-oriented classification aided by NDVI segmentation.Meanwhile,the vegetation information was also extracted by using the same object-oriented classification method without the application of NDVI.The extraction results were compared by the maximum likelihood supervised classification method.Satisfactory results were achieved by NDVI segmentation aided method with an overall accuracy of 87.69% and a KAPPA coefficient of 0.86.In contrast,the overall accuracy and Kappa coefficient of object-oriented classification method without NDVI were 82.69% and 0.80.The overall accuracy and Kappa coefficient of the maximum likelihood supervised classification were 71.92% and 0.67.The comparison indicated that object-oriented classification method could improve the classification accuracy.The application of NDVI in the segmentation could further improve the extraction accuracy of wetland vegetation.The method in this paper can achieve a more precise information extraction of the wetland vegetation.