基于分布差异的多时相SAR图像变化检测  

Multi-temporal SAR Image Change Detection Based on Distribution Divergence

作　　者：季艳[1,2] 胡蕾[1] 

机构地区：[1]北京航空航天大学计算机学院数字媒体实验室,北京100191 [2]北京遥感信息研究所,北京100011

出　　处：《遥感技术与应用》2010年第2期277-281,共5页Remote Sensing Technology and Application

摘　　要：结合SAR的后向散射特性和成像特点,利用交叉熵给出一种基于Edgeworth多项式的分布差异法检测不同时相的SAR图像变化区域。该方法采用Edgeworth多项式拟合SAR图像的密度分布,利用交叉熵计算两幅图像的差异指数,获得的差异图像能较好地反映图像变化情况,抑制了SAR图像固有的斑点噪声影响,采用CFAR分割差异图像得到的变化检测结果表明该方法具有高检测率、低虚警率的特点。According to the characteristic of SAR＇s backscatter and imagery feature,the paper presents a distribution divergence method calculating the cross-entropy of Edgeworth polynomial to detect the change area in different temporal SAR images.The method simulates the SAR images＇ probability distribution function with Edgeworth polynomial,and simulates the differentia index between two images with cross-entropy.The differentia image from differentia index reflects the images change,and reduces the influence of SAR images speckle noise.The detected results which is the different image segmentation results with CFAR demonstrate the method be of high detecting rate and low fault alarm rate.

关 键 词：合成孔径雷达(SAR) 分布差异法 变化检测 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]