卢丽琛　洪亮

摘 要：以武漢市东湖高新技术开发区部分区域为研究区，提出基于面向对象的高分辨率遥感影像建筑物变化检测法.利用BMI算法提取建筑物，利用CVA算法进行变化检测得到全部对象差异度，利用EM算法的贝叶斯阈值计算方法确定变化阈值.结果表明，基于面向对象的变化检测总体精度为89.48%，Kappa系数为0.86，优于基于像元的变化检测，为高分辨率遥感影像建筑物的变化检测提供了一种新的思维方式和方法.

关键词：高分辨率;变化检测;基于像元;面向对象

[中图分类号]P237   [文献标志码]A

Object-oriented Detection of Building Changes Based on HighSpatial Resolution Remote Sensing Image

LU Lichena，b，c，d，HONG Liang*a，b，c，d，e，f，g

（Yunnan Normal University，a.Faculty of Geography;b.GIS Technology Research Center of Resource and

Environment in Western China of Ministry of Education;c.Center for Geospatial Information Engineering

and Technology of Yunnan Province;d.Key Laboratory of Resources and Environmental Remote

Sensing for Universities in Yunnan;e.Center for Bay of Bengal Area Studies of Yunnan Normal

University;f.Center for Myanmar Studies of Yunnan Normal University;g.Center for Cambodia

Studies of Yunnan Normal University，Kunming 650500，China）

Abstract：Based on the object-oriented method of building change detection，a new method of building change detection based on high-resolution remote sensing image is proposed.BMI algorithm was used to extract buildings，CVA algorithm was used to detect changes to obtain all object differences，and the bayesian threshold calculation method of EM algorithm was used to determine the change threshold.The results show that the overall accuracy of object-based change detection is 89.48% and the Kappa coefficient is 0.86，which is superior to pixel-based change detection and provides a new way of thinking and method for the change detection of buildings with high-resolution remote sensing image.

Key words：high spatial resolution;change detection;pixel-based;object-oriented

近年来，随着遥感影像的空间分辨率不断提高和影像处理技术方法层出不穷，遥感影像变化检测技术取得很大突破，出现了许多新方法和实用性强的应用.在城镇区域的高分辨率遥感图像中，建筑物目标占据很大一部分，作为地物类别的主要内容，建筑物是城市地理数据库中最容易发生变化和最需要更新的部分，其更新较快，因此，开展对高分辨率遥感影像中的建筑物进行自动提取和变化检测研究具有重要的意义.最佳变化检测要结合具体的数据情况和实际应用目的来进行选择.高分辨率遥感影像的建筑物变化检测主要是对不同时间同一地区的两景或多景高分辨率遥感影像中的建筑物的变化情况进行检测，通过对比和分析，获取人们所需要的两时段建筑物的变化信息.[1-8]本文基于样本选择算法的贝叶斯阈值确定面向对象……