摘要
本文选取胶东半岛最具代表性的5个果品县(市)为研究区,以Landsat TM影像数据为分类影像,尝试提取果园信息。选用可以"无缝"融入多种辅助信息的决策树分类方法,综合NDVI、地形地貌和缨帽变换等多种辅助信息,利用年内物候变化最大的果园与背景地物的光谱差异,进行果园信息提取;利用SPOT影像以及野外考察资料作为检验样本进行精度验证。表明综合多种辅助信息,利用决策树分类法提取TM影像果园信息可行且准确性较高。
Decision tree classification is a kind of classification model which uses certain classification rules to gradually thin the research image.It has been widely used for information extraction from remote sensing images due to its goodness of intuitive and high efficiency.Jiaodong Peninsula is one of the most famous areas in China for the production of fruits;therefore,it is very significant to monitor the distribution of orchards.In this paper,the decision tree classification was used to extract the area of orchard in Jiaodong Peninsula.Specifically,Landsat5 TM image（path120 row034,October24,2005） was available and five most representative cities（Penglai,Longkou,Laizhou,Qixia,Zhaoyuan） were selected as the study area.It turned out that the decision tree classification had satisfactory performance,the classification results were acceptable and could be used as the original inputs for related researches.
出处
《测绘科学》
CSCD
北大核心
2012年第4期57-60,共4页
Science of Surveying and Mapping
基金
国家自然科学基金项目(40801016)
中国科学院知识创新工程重要方向项目(kzcx2-yw-224)
关键词
TM影像
决策树
果园
信息提取
胶东半岛
TM imagery
decision tree
orchard
information extraction
Jiaodong Peninsula
