基于MMSE-T的合成孔径雷达图像超分辨率重建

MMSE-T based super-resolution reconstruction of synthetic aperture radar image

针对雷达目标图像,提出一种基于阈值最小均方误差(MMSE-T)的超分辨率重建方法,并对其性能进行了分析、比较和评估.介绍和分析了雷达成像模型及常用的超分辨方法.以及MMSE-T改进算法及其具体实现方法.以MSTAR合成孔径雷达(SAR)实测图像为例,给出其超分辨结果,同时基于输出信噪比(SNR)指标,对其性能进行了比较与评估.实验表明:MMSE-T超分辨率方法在无须事先已知原始场景先验知识的情况下,可实现对原始场景的准确重建,同时具有较好的噪声抑制作用,可用于高分辨率一维距离像、合成孔径雷达、逆合成孔径雷达及实波束成像等雷达图像目标信息的开发.

A radar image super-resolution reconstruction approach based on thresholded minimum mean-square error（MMSE-T） technique was given,and its super-resolution performance was analyzed,compared and assessed.Radar imaging model and several common super-resolution algorithms were introduced.Then an improved MMSE-T super-resolution algorithm and its realization method were described.The algorithm was demonstrated using MSTAR synthetic aperture rodar（SAR） measured images,and its performance was assessed and compared by the index-the output signal-to-noise ratio（SNR）.The experimental results indicate that the MMSE-T approach can accurately reconstruct the original scene without prior knowledge,and has good effect of noise suppression.The method can be applied to exploit target information from the radar images produced by high-resolution range profile,SAR inverse SAR or real beam imaging radar.