面向对象的那曲WorldView-2影像的分类  

Object-oriented Image Classification of Naqu Using WorldView-2 Image

作　　者：倪卓娅[1,2] 甘甫平[2] 周萍[1] 霍红元[1] 

机构地区：[1]中国地质大学地球科学与资源学院,北京100083 [2]中国国土资源航空物探遥感中心对地观测技术工程实验室,北京100083

出　　处：《遥感信息》2011年第6期114-118,共5页Remote Sensing Information

基　　金：国家863计划:2006AA06A208;中国地质调查局地局大调查项目1212011087113资助

摘　　要：随着WorldView-2卫星的发射,首次出现了8波段多光谱高分辨率商业卫星。高分辨率影像包含丰富的空间结构信息和地理特征信息,采用传统的基于像元的影像分析方法精度明显达不到要求。为了提高精度,本文以西藏那曲为研究区,利用面向对象的方法对那曲的WorldView-2影像进行分类,采用模糊分类法,利用所分地物的特征确定类别,针对西部地区干旱河流的特殊情况,提出了一种基于地物空间轮廓特征的分类方法,分类结果较好。The WorldView-2 satellite is the first 8-band multi-spectral high-resolution commercial satellite.High-resolution image contains a wealthy information on spatial structure and geographic features,so the precision of traditional classification method based on pixel cannot meet the demand.In this paper,taking Naqu city of Tibet autonomous region as a study area and in order to improve the classification accuracy,the object-oriented method is used to do Naqu WorldView-2 image classification,combining with fuzzy classification,and categorize the target objects according to their surface features.Considering the special situation that the western rivers are always dry and frozen in February,a method based on object contour and space features is brought up and the results of classification are good.

关 键 词：WorldView-2 高分辨率影像 面向对象 ECOGNITION 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]