电子学报 ›› 2021, Vol. 49 ›› Issue (12): 2339-2348.DOI: 10.12263/DZXB.20200481

• 学术论文 • 上一篇    下一篇

距离-多普勒-频带域3D-AWP-MRF分类辅助的SAR-GMTI杂波抑制方法

韩超垒1, 杨志伟1, 张庆君2, 廖桂生1, 何鹏远1   

  1. 1.西安电子科技大学雷达信号处理国家重点实验室,陕西 西安 710071
    2.中国空间技术研究院,北京 100094
  • 收稿日期:2020-05-21 修回日期:2021-08-08 出版日期:2021-12-25 发布日期:2021-12-25
  • 作者简介:韩超垒 男,1993年生,河南人,西安电子科技大学博士研究生.主要研究方向为SAR成像技术、机载/星载地面动目标检测和多维信号处理.
    杨志伟(通信作者) 男,1980年生,四川人,西安电子科技大学教授,博士生导师.主要研究方向为阵列信号处理、空-时-极化自适应处理、机载/星载地面动目标检测、多维信号处理和稀疏信号处理.E-mail:yangzw@xidian.edu.cn
    张庆君 男,1969年生,江苏人,卫星设计总师.主要研究方向为卫星总体设计及微波遥感技术.
    廖桂生 男,1963年生,广西人,西安电子科技大学教授,博士生导师.主要研究方向为阵列信号处理、空-时自适应处理、机载/星载地面动目标检测、分布式卫星系统设计.
  • 基金资助:
    国家自然科学基金(62071481);国家青年基金(61801373);上海航天科技创新基金资助项目(SAST2018043)

A Multichannel SAR-GMTI Clutter Suppression Method Assisted by Range-Doppler-Band 3D-AWP-MRF Classification

HAN Chao-lei1, YANG Zhi-wei1, ZHANG Qing-jun2, LIAO Gui-sheng1, HE Peng-yuan1   

  1. 1.National Laboratory of Radar Signal Processing,Xidian University,Xi’an,Shaanxi 710071,China
    2.China Academy of Space Technology,Beijing 100094,China
  • Received:2020-05-21 Revised:2021-08-08 Online:2021-12-25 Published:2021-12-25

摘要:

针对复杂地理场景下非均匀杂波造成多通道SAR-GMTI(Synthetic Aperture Radar Ground Moving Target Indicator,合成孔径雷达动目标指示)系统杂波抑制性能下降的问题,本文提出一种距离-多普勒-频带域三维自适应加权惩罚马尔科夫场(Three Dimension Adaptive Weighted Penalty Markov Random Field,3D-AWP-MRF)分类辅助的SAR-GMTI杂波抑制方法. 利用地物类型在距离-多普勒-频带三维SAR图像空间的马尔科夫性,构造了包含空间距离、类间Fisher距离、局部粗糙度距离和梯度方向距离的自适应加权惩罚函数. 在贝叶斯框架下利用条件迭代模型(Iterated Conditional Mode,ICM)算法求解距离-多普勒单元所属地物类型的最大化后验概率对多频带SAR图像进行分类,接着利用图像形态学操作对图像分类结果进行区域提取. 最后,对每个闭合区域分别估计杂波协方差矩阵,并进行自适应杂波抑制处理. 相比于传统方法,本文方法在非均匀杂波环境下不仅能提高起伏区域强杂波10~15dB的强杂波抑制能力,还可以减少平坦区域慢速运动目标约2.5dB的输出能量损失.

关键词: SAR-GMTI, 杂波抑制, 多频带, 马尔科夫, 杂波分类, 自适应惩罚加权

Abstract:

To address the problem that the clutter suppression performance of a multichannel synthetic aperture radar ground moving target indicator(SAR-GMTI) system is reduced due to non-uniform clutter in a complex geographical scene, we propose a multichannel SAR-GMTI clutter suppression method assisted by Range-Doppler-Band three dimension adaptive weighted penalty markov random field(3D-AWP-MRF) classification. Firstly, an adaptive weighted penalty function is constructed using the markov properties of the feature types in the Range-Doppler-Band 3D SAR image domain. Spatial distance, inter-class Fisher distance, local roughness distance, and gradient direction distance are included. Subsequently, in the Bayesian framework, iterated conditional mode(ICM) algorithm is used to solve the maximum posterior probability of the feature type of the Range-Doppler unit, so as to accurately classify the multiband SAR images. Then, the image classification results are obtained by image morphology operation. Finally, the clutter covariance matrix is estimated separately for each closed region, and adaptive clutter suppression processing is performed. Compared with the traditional method, the proposed method not only can improve the suppression performance about 10~15 dB for the strong clutter in undulating areas, but also reduce the output signal power loss about 2.5dB of the moving targets in flat areas.

Key words: SAR-GMTI, clutter suppression, multiband, Markov, clutter classification, Adaptive penalty weighting

中图分类号: