摘要
利用分层分类思想,对黑龙江省塔河县森林类型使用多期TM遥感数据进行分类。根据研究区域实际情况,制定二级分类系统。使用7个时相的TM原始波段数据及其计算的植被指数表达森林类型的光谱特征,采用抽样方法选取训练样本并进行分析,统计不同森林类型的TM原始波段和多种植被指数在各时相上的差异,得到差异较大的波段和植被指数作为分类特征,采用最大似然法和决策树分类方法对森林类型进行分类。研究结果表明,决策树分类方法对优势树种组的分类精度最高,达到76.85%,可以满足实际应用的要求。
The forest types in Tahe county of Heilongjiang province were classified using multi-temporal TM images based on stra- tum sampling. Combined with the characteristics of remote sensing technology and the actual situation in the study area, a two-level classification system was designed for the research area. Seven-temporal basic TM data and the calculated vegetation indices were used to express the spectrum features of the forest types, and the random sampling and systematic sampling were used to choose and analyze the training samples in all levels of classification. The differences of TM original bands in different forest types and multiple vegetation indices in different temporal were summarized and the proper classification features were obtained to classify the TM images. Finally, the maximum likelihood estimate and decision tree classification methods were used to classify the forest types. The results showed that the classification accuracy for the dominant species using the decision tree classification method is the highest, which can be up to 76. 85% and meet the requirements of actual application.
出处
《森林工程》
2013年第2期14-20,共7页
Forest Engineering
基金
国家高技术研究发展计划资助(2012AA102001)
国防科工局专项资助(E0305/1112/01/01)
关键词
分类
多时相
信息提取
决策树
时相间植被指数
classification
multi-temporal
information extraction
decision tree
vegetation indices between temporal
