首页 | 本学科首页   官方微博 | 高级检索  
     检索      

高分辨率遥感图像均值调整法分割技术研究
引用本文:林辉,莫登奎,熊育久,孙华.高分辨率遥感图像均值调整法分割技术研究[J].中南林业科技大学学报,2006,26(4):85-88.
作者姓名:林辉  莫登奎  熊育久  孙华
作者单位:中南林业科技大学资源与环境学院 湖南长沙410004
基金项目:国家自然科学基金;湖南省杰出青年科学基金;湖南省教育厅青年基金
摘    要:遥感图像分割是将以像元为基础的图像转化为以对象为基础的过程,因此,遥感图像分割是遥感图像高级分析的基础.对于中、低分辨率遥感图像来说,单个像元的面积较大,且混合像元现象严重,图像分割会产生较大的误差.高分辨率遥感图像单个像元的面积较小,分割后图像上单个对象内含若干像元,便于分析和提取信息.以高分辨率遥感图像Q u ickB ird和IKONO S为研究对象,采用均值调整法对图像进行分割和精度检验.结果表明:采用均值调整法进行高分辨率遥感图像分割具有较好的效果,图像分割的速度和精度均较高.

关 键 词:遥感  分割  均值调整法  模型  精度
文章编号:1000-2502(2006)04-0085-04
修稿时间:2006年1月22日

The Method of Segmenting High-resolution Remote Sensing Images Based on Mean-shift
LIN Hui,MO Deng-kui,XIONG Yu-jiu,SUN Hua.The Method of Segmenting High-resolution Remote Sensing Images Based on Mean-shift[J].Journal of Central South Forestry University,2006,26(4):85-88.
Authors:LIN Hui  MO Deng-kui  XIONG Yu-jiu  SUN Hua
Abstract:Image segmentation is a process of changing pixel-based images into object-based ones;so in remote sensing,image segmentation is the base of high-level image analysis.Image segmentation brings errors in low or medium-resolution remote sensing images because a single pixel covers a larger area and there are a lot of mixed pixels.On the contrary,in a high-resolution remote sensing image,a pixel covers a smaller area and there are less mixed pixels;so a single object on a segmented image includes a number of pixels,and thus makes it possible to analyze and extract information.In this paper,QuickBird image was segmented by mean-shift and its accuracy was checked.The result shows that mean-shift is a good method of segmentation to high-resolution remote sensing images;both the velocity and accuracy of segmentation are satisfied.
Keywords:remote sensing  segmentation  mean-shift  models  accuracy
本文献已被 CNKI 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号