SAR图像配准的二元Rayleigh分布互信息方法  

Bivariate Rayleigh Distribution and its Mutual Information for SAR Image Registration

作　　者：肖星[1] 田铮[1,2] 冷成财[1] 

机构地区：[1]西北工业大学应用数学系,西安710072 [2]中国科学院遥感应用研究所国家重点实验室,北京100101

出　　处：《工程数学学报》2011年第2期157-164,共8页Chinese Journal of Engineering Mathematics

基　　金：国家自然科学基金(60972150;10926197);航空科学基金(20081913006)~~

摘　　要：合成孔径雷达(Synthetic Aperture Radar,简称SAR)图像中乘性斑点噪声的存在,使得传统的互信息配准方法中的插值假象更为严重.本文把单视SAR幅值图像同质区域服从Rayleigh分布这一先验信息引入互信息配准方法中,对同源单视SAR幅值图像进行配准.以合成图像和真实图像的试验结果和分析证实了方法的可行性,说明新方法优于传统直方图估计联合分布的互信息方法.Due to the speckle noise of Synthetic Aperture Radar （SAR） image which is determined by the SAR imaging mechanism,the interpolation artifacts of mutual information based on image registration become important.This paper combines the prior information that the homogenous regions of the single-look SAR amplitude image follow the Rayleigh distribution with the mutual information,and registers two single-look SAR amplitude images by the bivariate Rayleigh distribution and its mutual information.The experiments of synthetic images and real SAR images show the efficacy and advantage of this method when compared with the traditional mutual information method.

关 键 词：单视幅值 SAR 图像配准 二元Rayleigh分布 互信息 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]