基于互信息和多智能体优化的合成孔径雷达图像配准  

Mutual information based multi-agent optimization for SAR image registration

作　　者：王凌霞[1] 颜学颖[1] 

机构地区：[1]西安电子科技大学电子工程学院智能感知与图像理解教育部重点实验室,陕西西安710071

出　　处：《电波科学学报》2014年第6期1094-1097,1146,共5页Chinese Journal of Radio Science

基　　金：国家自然科学基金(No.61072106);国家教育部博士点基金(No.20100203120005);高等学校学科创新引智计划(111计划)(No.B07048)

摘　　要：提出了一种基于互信息的多智能体优化方法,并将其用于合成孔径雷达(Synthetic Aperture Radar,SAR)图像的配准.采用归一化互信息作为度量准则,直接利用图像的灰度数据进行配准,没有假设图像间灰度值的线性关系,不需要对图像做分割、特征提取等预处理,并利用多智能体优化算法寻找最佳匹配参数,以得到最优的配准结果.实验表明,该算法在SAR图像配准方面有更好的普适性和更高的配准精度.Mutual information based multi-agent optimization algorithm is proposed for Synthetic Aperture Radar（SAR） image registration. This algorithm adopts normalized mutual information as the measure criterion, which is directly based on the brightness information of images without the assumption of linear relationship between brightness values of images and the preprocessing steps such as image segmentation and feature ex- traction. A mutual information based multi-agent optimization algorithm is carefully de- signed to obtain final registration results. Experiments on real SAR images validate the effectiveness and accuracy of the proposed algorithm.

关 键 词：SAR图像配准 多智能体系统 互信息 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]