一种综合孔径辐射计贝叶斯反演法  

A Bayesian inversion method for aperture synthesis radiometers

作　　者：何方敏[1] 李青侠[1] 冷毅[2] 朱耀庭[1] 

机构地区：[1]华中科技大学电子与信息工程系,湖北武汉430074 [2]空军雷达学院信息对抗系,湖北武汉430019

出　　处：《华中科技大学学报（自然科学版）》2010年第8期1-4,共4页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(60705018)

摘　　要：将贝叶斯反演理论应用于综合孔径辐射计图像反演,其基本思想是将亮温分布和可见度视为随机变量,将图像反演转换为贝叶斯推断问题.基于这一思想,分析了亮温分布的先验信息和可见度的统计特性.鉴于热辐射信号具有高斯特性,建立了亮温分布的高斯先验模型,并采用期望最大算法估计该先验模型的未知参数.可见度采用多元高斯随机变量表示,并考虑了可见度的互相关性.贝叶斯反演法能够充分利用可见度和先验信息的统计特性.仿真和实验结果表明:贝叶斯反演法能有效提高综合孔径辐射计的成像性能.Bayesian inversion methods, whose principle is to treat the unknown brightness temperature distribution and the visibilities as random variables and to recast the image reconstruction of ASRs as Bayesian inference, was used in the image reconstruction of aperture synthesis radiometers （ASRs）. The statistical properties of the visibilities and the prior information of brightness temperature are investigated. Thermal radiation having Gaussian property, a Gaussian prior model was constructed by an unknown parameter, which can be estimated by the expectation maximization （EM） algorithm. The visibilities were modeled as multivariate Gaussian variables, and the cross-correlations of the visibilities were deduced. Bayesian inversion method can seamlessly account for the statistical properties of visibilities and the prior information. The simulation and experiment results show that Bayesian inversion approach can improve the imaging performance of ASRs. This approach can be applied to highaccuracy image reconstruction of ASRs.

关 键 词：综合孔径 辐射计 反问题 贝叶斯反演法 先验信息 统计特性 

分 类 号：TN911.23[电子电信—通信与信息系统]