基于多地表特征参数的遥感影像分类研究  

Classification of Remote Sensing Image Based on Multi-parameters of Land Surface

作　　者：曹丽琴[1] 李平湘[2] 张良培[2] 岑奕[3] 

机构地区：[1]武汉大学印刷与包装系,湖北武汉430079 [2]武汉大学测绘遥感信息工程国家重点实验室,湖北武汉430079 [3]长江水利委员会长江科学院水土保持所,湖北武汉430079

出　　处：《遥感技术与应用》2010年第1期38-44,共7页Remote Sensing Technology and Application

基　　金：国家"973计划"资助项目(2009CB723905);国家"863"计划课题项目(2007AA12Z148;2007AA12Z181;2009AA12Z114);国家自然科学基金项目(40771139;40523005);"极地测绘科学国家测绘局重点实验"开放基金联合资助

摘　　要：地表特征是反映地表信息的重要参数,是了解地表时空多变信息的定量要素。提出基于多地表特征参数的遥感影像分类方法,并利用武汉市的Landsat ETM+影像为例进行试验。试验选择通用植被指数(VIUPD)、地表温度和纹理特征等多地表特征参数,在考虑光谱特征和空间信息的前提下,结合分层思想的决策树方法,对遥感影像进行分类。结果证明利用多地表特征参数的决策树分类方法与传统的基于光谱反射率特征的决策树分类方法和SVM分类方法相比较,分类精度有了明显的提高。The land cover characteristic is the important parameter of land surface information,which is the essential factor to understand the change temporal-spatial information of land surface.In this paper,the method of classification based on multi-parameters of land surface was proposed.The data of the experimenter was Landsat ETM＋ image of Wuhan.The Vegetation Index of Universal Pattern Decomposition（VIUPD）,land surface temperature and the character of texture were carried.According to the method of Decision Tree,the image was classified in the consideration of the spectral and spatial information.The results demonstrated that the method of classification based on multi-parameters of land surface was superior to traditional Decision tree based on the spectral reflectance and Supported Vector Machine（SVM）.The accuracy of classification based on multi-parameters of land surface was higher than that of the other algorithms.

关 键 词：多地表特征参数 影像分类 决策树 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]