基于局部互相关的遥感图像变化检测技术  

A Local Neighborhood Correlation Method to Change Detection Using Multitemporal Remote Sensing Images

作　　者：刘攀龙[1] 

机构地区：[1]91404部队,秦皇岛066000

出　　处：《舰船电子工程》2016年第7期150-153,共4页Ship Electronic Engineering

摘　　要：针对遥感图像变化检测的全局假设与实际不一致问题,论文提出了基于局部邻域的互相关的遥感图像变化检测算法。该算法首先运用区域划分的思想,以像素灰度级为单位,对图像进行分组处理;其次以多时相遥感图像的差值图像为观察场,以像素所属类别为标号场,建立二者互相关的马尔科夫随机场算法;最后采用两段式步骤,对算法进行迭代处理。实验结果表明在遥感图像变化检测算法中,该方法具有较高的检测精度。To solve the homogeneity between simple Markov Random Field（MRF）assumptions with its actual situation,in this paper a method to change detection based on local neighborhood correlation MRF is proposed.For this method,firstly,the regional division strategy is applied to process images based on pixel’s gray.Secondly,the difference image is taken as observation field,and the obtainedclasses is taken as labeling field,in which the Correlation Markov Field（CMRF）is built.At last,the two-stage iteration algorithm is used to solve the CMRF method.The experimental shows that this method has high detection accuracy.

关 键 词：多时相遥感图像 互相关马尔科夫随机场 局部邻域 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]