摘要
在超宽带穿墙雷达成像应用中,由于目标成像空间的稀疏性表示可以采用压缩感知在较少数据采集下获得高分辨、低旁瓣的成像结果。然而,这些稀疏成像方法要求真实目标必须位于预设的网格点上才能保证较好的成像质量,否则会出现严重的虚假目标像。本文提出一种基于修正贝叶斯压缩感知的偏离网格目标稀疏成像方法。该方法对偏离网格目标的感知矩阵在预设网格点进行一阶泰勒级数展开,把真实目标与网格点之间的偏移量视为稀疏贝叶斯模型的参数,然后通过构造的联合概率密度函数利用最大期望方法联合估计目标散射系数和偏移量,保证了目标真实像的准确恢复。仿真和实验结果表明,该方法通过校正目标偏离网格引起的模型误差有效改善了传统稀疏重建方法的成像质量。
UWB through-the-wall radar imaging (TWRI) based on compressed sensing (CS) can decrease the sampling time and data volume while generating high-resolution low sidelobe images. Good reconstruction performance is based on the assumption that all the tar- gets are exactly located at the pre-discretized grid points ; otherwise, the performance would significantly degrade and lead to serious false target image. In this paper, a spare imaging method based on modified bayesian compressive sensing (MBCS) in TWR is proposed for off-grid target. The first-order Taylor series expansion is implemented for the sensing matrix of the off-grid target in the pre-discretized grid points. The difference between the actual target positions and the pre-discretized points is regarded as the sparse Bayesian model pa- rameter. A joint probability density function is constructed, and the expectation-maximization (EM) algorithm is used to estimate the target scattering coefficients and offsets, and insure the accurate localization of the target image. The simulation and experiment results show that the proposed method can correct the model error caused by off-grid target, effectively improve the imaging quality and is better than conventional sparse reconstruction method,
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2015年第4期743-748,共6页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金(61461012)
广西区自然科学基金(2013GXNSFAA019329
2013GXNSFAA 019004)
广西无线宽带通信与信号处理重点实验室项目(GXKL0614106)资助项目
关键词
穿墙雷达稀疏成像
偏离网格问题
修正贝叶斯压缩感知
through-the-wall radar sparse imaging
off-grid problem
modified Bayesian compressive sensing
作者简介
晋良念,1998年和2003年于桂林电子科技大学获得学士学位和硕士学位,2012年于西安电子科技大学科技获得博士学位,主要研究方向自适应信号处理,超宽带雷达隐藏目标成像与识别。E—mail:jinglingling5653@sina.com.en;钱玉彬,2013年于滁州学院获得学士学位,现为桂林电子科技大学硕士研究生,主要研究方向为超宽带穿墙雷达稀疏成像。E-mail:qianyubin1990@sina.com