TM影像决策树分类中的影响因素研究

以云南省西双版纳州一景TM影像为例，分析了影响分类回归树方法的主要因素。结果表明在其他因素均一致的情况下，训练数据如果使用涵盖各类别的外业调查数据比使用系统布设的训练数据分类精度更高，并且多种参数波段的选择也会有效地提高分类的精度。...

Full description

Saved in:
Bibliographic Details
Published in林业科学研究 Vol. 27; no. 1; pp. 1 - 5
Main Author 张连华 庞勇 岳彩荣 李增元 范应龙 谭炳香 车学俭
Format Journal Article
LanguageChinese
Published 中国林业科学研究院资源信息研究所,北京 100091 2014
西南林业大学林学院,云南昆明 650224%中国林业科学研究院资源信息研究所,北京,100091%西南林业大学林学院,云南昆明,650224
Subjects
TM影像
决策树分类
植被指数
波段组合
训练数据
vegetation index
训练数据
decision tree
TM影像
波段组合
training data
植被指数
bands combination
TM image
决策树分类
Online AccessGet full text
ISSN1001-1498

Cover

More Information
Summary:以云南省西双版纳州一景TM影像为例，分析了影响分类回归树方法的主要因素。结果表明在其他因素均一致的情况下，训练数据如果使用涵盖各类别的外业调查数据比使用系统布设的训练数据分类精度更高，并且多种参数波段的选择也会有效地提高分类的精度。
Bibliography:11-1221/S
ZHANG Lian-hua, PANG Yong, YUE Cai-rong, LI Zeng-yuan, FAN Ying-long , TAN Bing-xiang , CHE Xue-jian （ 1. Research Institute of Forest Resource Information Techniques, Chinese Academy of Forestry, Beijing 100091, China; 2. College of Forest, Southwest Forestry University, Kunming 650224, Yunnan, China）
Taking one scene TM image of Xishuangbanna of Yunnan as an example, the main factors affecting the classification and regression tree method were analyzed. The results show that in the parameters under identical cir- cumstances, the training data has higher classification accuracy if the field investigation data covering all the classi- fication data were used rather than the system layout data. It also shows that selecting various bands of parameters can also improve the precision of classification effectively.
decision tree; TM image; training data; vegetation index; bands combination
ISSN:1001-1498