基于多时相遥感影像变化检测并联合GIS系统的地形图更新方法研究  

A change detection algorithm for Man-made objects based on Multi-temporal remote sensing images join GIS for map updating

作　　者：贾有良 

机构地区：[1]武汉大学,武汉430079 [2]天津市测绘院,天津300381

出　　处：《测绘科学》2008年第S2期49-51,共3页Science of Surveying and Mapping

基　　金：国家测绘局地理信息工程重点实验室;国家测绘局青年学术技术带头人项目资助

摘　　要：多时相遥感图像变化检测在国民经济和国防建设领域有着广泛的应用需求。通过分析同一地域不同时相的遥感图像,变化检测提供地物发生变化的信息,用于城市规划、资源和环境监测、战场态势分析以及打击效果评估等。迄今为止,众多学者己经提出了多种变化检测的方法,这些方法按照信息处理的层次可分为基于像素,基于特征和基于目标的变化检测。本文将针对城市变化检测的需要,利用SPOT5卫星数据主要研究了分析和测试了目前主要的变化检测方法在城市建筑和用地变化的应用,并试验了基于Mean-Shift图像分割的变化检测方法。试验证明这一方法是有效的。Analysis the Multi-temporal Remote Sensing Images to check out the change of ground target in different image which cover the correspond area,this information will be useful in city planning,resource and environment monitor,and so on.So far,a lot of method had been explored for the change detection in multi-temporal image by many scholars,according to category of information, they are belonged to pixel-based,feature-based and target-based change detection method.This paper will analysis and test part of them in the practice for city building and land using changes detection,and presents one method based on mean-shift algorithm, show the efficiency in the work of change detection.

关 键 词：多时相 遥感影像 变化检测 SUSAN算子 MEAN-SHIFT算法 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]