基于Mean-shift影像分割算法的TM影像分类及城市扩展研究  

Study of TM Image Classification Based on Image Segmentation of Mean-Shift Algorithm and Its Application in Urban Sprawl

作　　者：史晓霞[1,2] 李京[2,3] 周冠华[1,2] 赵静[1,2] 

机构地区：[1]北京师范大学资源学院,北京100875 [2]地表过程与资源生态国家重点实验室(北京师范大学),北京100875 [3]民政部/教育部减灾与应急管理研究院(北京师范大学),北京100875

出　　处：《地理与地理信息科学》2009年第1期110-112,共3页Geography and Geo-Information Science

基　　金：国际科技合作计划专项项目(2007DFA20640);国家科技支撑计划课题(2006BAJ05A01;2006BAJ09B06;2006BAJ09B03);国家自然科学基金(40701114);高等学校博士学科点专项科研基金(20070027017);北京市自然科学基金(8082015);江苏省资源环境信息工程重点实验室开放基金(20080102)

摘　　要：以马来西亚雪兰莪州及首都吉隆坡为例,将Mean-shift算法的影像分割方法应用于研究区的TM影像分类中,较传统方法的分类精度更高,可满足应用需求。依据该分类结果对研究区的空间城市化特征进行分析,发现近期研究区一直处于高速扩展阶段,外延式扩展和填充式扩展并行发展成为该研究区城市扩展的主导模式。Image segmentation is one of the methods of image classification. In this paper,Selangor Malaysia and Kuala Lumpur, the capital of Malaysia, were taken as an example and the Mean-shift algorithm, which is one of the methods of image segmentation, was employed to do Landsat/TM image classification. Then the classification result was used to do urban sprawl characteristic analysis. The classification result got from the image segmentation method based on Mean-shift algorithm shows that the precision is very high. The process of urban sprawl belongs to the combination of extension sprawl model and filling sprawl model. The result shows that TM image classification, based on the image segmentation of Mean-shift algorithm provides a new approach with high accuracy for TM image classification.

关 键 词：MEAN-SHIFT算法 影像分割 城市扩展 

分 类 号：TP722.4[自动化与计算机技术—检测技术与自动化装置]