基于GPD的稳健贝叶斯压缩感知ISAR成像方法  

Robust Bayesian Compressed Sensing ISAR Imaging Based on Generalized Pareto Distribution

作　　者：夏朝禹 高瑜翔[1,2] 黄坤超 包兆华[1,2] 郭春妮 王震 XIA Chaoyu;GAO Yuxiang;HUANG Kunchao;BAO Zhaohua;GUO Chunni;WANG Zhen(College of Communication Engineering,Chengdu University of Information Technology,Chengdu 610225,China;Meteorological Information and Signal Processing Key Laboratory of SichuanHigter Education Institutes,Chengdu 610225,China;Southwest China Institute of Electronic Technology,Chengdu 610036,China;Aviation Military Representative Office of Army Armament Department in Chengdu Region,Chengdu 610036,China)

机构地区：[1]成都信息工程大学通信工程学院,成都610225 [2]气象信息与信号处理四川省高校重点实验室,成都610225 [3]中国西南电子技术研究所,成都610036 [4]陆军装备部驻成都地区航空军代室,成都610036

出　　处：《电讯技术》2020年第5期585-590,共6页Telecommunication Engineering

摘　　要：现有贝叶斯压缩感知(Bayesian Compressed Sensing,BCS)-逆合成孔径雷达(Inverse Synthetic Aperture Radar,ISAR)成像算法中先验分布模型不能很好地满足可压缩性,导致成像精度随脉冲数目的减小、高斯噪声的增强而急剧下降。为此,提出了一种基于广义Pareto分布改进BCS成像方法(Improving BCS imaging based on GPD,IGPCS)。该方法主要在BCS框架下利用广义Pareto先验分布替代传统的广义Gaussian先验分布,以增强模拟信号的稀疏先验和可压缩性。进一步地,为了克服后验概率模型计算困难等问题,采用最大后验(Maximum A Posteriori,MAP)方法对超参数进行估计。通过对Mig-25小型飞机的ISAR模拟实验表明,与传统方法相比,IGPCS方法能够获取极高的成像精度,并且对低脉冲数、强高斯噪声环境具有较强的鲁棒性。The prior distribution model in the existing Bayesian compressed sensing inverse synthetic aperture radar( BCS-ISAR) imaging algorithm cannot well meet the compressibility,as a result,the imaging accuracy decreases sharply with the reduction of pulse number and the enhancement of Gaussian noise.In this paper,an improved BCS imaging method based on generalized Pareto distribution( GPD) called IGPCS is proposed.This method mainly uses GPD to replace the general Gaussian distribution( GGD) in the framework of BCS to enhance sparse prior and compressibility. Furthermore,the maximum a posteriori( MAP)method is used to estimate the hyper parameters and overcome the difficulties in calculating the posterior probability model.The ISAR simulation experiment of Mig-25 small aircraft shows that IGPCS method can obtain extremely high imaging accuracy compared with traditional method,and it has strong robustness to low pulse number as well as strong Gaussian noise environment.

关 键 词：逆合成孔径雷达 成像方法 贝叶斯压缩感知 广义Pareto先验分布 

分 类 号：TN957[电子电信—信号与信息处理] TN985[电子电信—信息与通信工程]