不同时相城市QuickBird和WorldView-2植被分类  

Recognition Urban Vegetation Types Based on QuickBird and WorldView-2 Images in Different Time Phase

作　　者：刘怀鹏 安慧君[2] LIU Huai-peng;AN Hui-jun(School of Land and Tourism,Luoyang Normal University,Luoyang 471934,China;College of forestry,Inner Mongolia Agricultural University,Hohhot 010019,China)

机构地区：[1]洛阳师范学院国土与旅游学院,河南洛阳471934 [2]内蒙古农业大学林学院,内蒙古呼和浩特010019

出　　处：《数学的实践与认识》2018年第20期43-49,共7页Mathematics in Practice and Theory

基　　金：中意智慧城市合作研究室项目(2016YFE0104600);内蒙古自然科学基金项目(2015MS0341)

摘　　要：影像植被分类中数据源的时相因素会对分类结果产生重要影响．研究以2月份和8月份成像的QuickBird、World View-2为数据源，利用最大似然分类法将两类影像分为针叶树、阔叶树和草类3个类别，分析植被生长旺盛季与非旺盛季数据在植被分类中的有效性．实验结果表明：QuickBird影像分类的总体精度为96．6667%，Kappa系数0．9500，World View-2分类的总体精度为93．9871％，Kappa系数0．9098，均取得较好的分类结果；QuickBird分类的总体精度较WorldView-2高2．6796％，Kappa系数高0．0402，Quick Bird的分类效果优于World View-2．研究结果表明适宜时相数据源的选择更加有利于城市植被类型的识别．The temporal factor of the data source has an important influence on the classifica- tion results in image vegetation classification. In this study, QuickBird and WorldView-2 were used as data sources which imaged in February and August, we used Maximum likelihood clas- sification divided these images into conifer tree, broad-leaf tree and grasses, and analyzed the validity of vegetation growth and non vigorous season data in vegetation classification. Exper- imental results showed： The overall accuracy of QuickBird image classification was 96.6667%, Kappa coefficient was 0.9500, the overall accuracy of WorldView-2 classification was 93.9871%, Kappa coefficient was 0.9098, both of them obtained a good classification results; the overall accuracy of QuickBird classification was higher 2.6796% than used WorldView-2, Kappa coef- ficient was higher 0.0402, the classification effect of QuickBird was better than WorldViewo2. It is showed that choose appropriate time phase image will more conducive to identification urban vegetation types.

关 键 词：QUICKBIRD World View-2 城市 时相 植被分类 

分 类 号：Q948[生物学—植物学] TP79[自动化与计算机技术—检测技术与自动化装置]