电子学报 ›› 2021, Vol. 49 ›› Issue (7): 1279-1285.DOI: 10.12263/DZXB.20201277

• 学术论文 • 上一篇    下一篇

超视距雷达中距离-多普勒图的瞬态干扰自动识别方法

罗忠涛1, 夏杭1, 卢琨2, 何子述3   

  1. 1.重庆邮电大学通信与信息工程学院, 重庆 400065
    2.南京电子技术研究所, 江苏 南京 210013
    3.电子科技大学信息与通信工程学院, 四川 成都 611731
  • 收稿日期:2020-11-12 修回日期:2020-12-09 出版日期:2021-07-25
    • 作者简介:
    • 罗忠涛 男,1984年生于四川隆昌.重庆邮电大学通信与信息工程学院副教授.研究方向为统计信号处理、数字图像处理与机器学习.E‑mail:luozt@cqupt.edu.cn
      夏 杭 男,1994年生于贵州安顺.重庆邮电大学通信与信息工程学院硕士研究生.研究方向为雷达信号处理、数字图像处理与机器学习.E‑mail: 1790095607@qq.com
      卢 琨 男,1977年生于广西桂林.南京电子技术研究所研究员级高级工程师.研究方向为超视距雷达系统设计与信息处理. E‑mail: mimimomoba@gmail.com
      何子述 男,1962年生于四川新繁.电子科技大学信息与通信工程学院教授.研究方向为雷达信号处理与阵列信号处理.E‑mail:zshe@uestc.edu.cn
    • 基金资助:
    • 国家自然科学基金 (61701067)

Automatic Recognition of Transient Interference in the Range‑Doppler Map for Over‑the‑Horizon Radar

LUO Zhong‑tao1, XIA Hang1, LU Kun2, HE Zi‑shu3   

  1. 1.School of Communication and Information Engineering, Chongqing University of Posts and Telecommunications, Chongqing 400065, China
    2.Nanjing Research Institute of Electronics Technology, Nanjing, Jiangsu 210013, China
    3.School of Information and Communication Engineering, University of Electronic Science and Technology of China, Chengdu, Sichuan 611731, China
  • Received:2020-11-12 Revised:2020-12-09 Online:2021-07-25 Published:2021-08-11
    • Supported by:
    • National Natural Science Foundation of China (61701067)

摘要:

为提高瞬态干扰处理的稳健性,超视距雷达可采用先识别后抑制的思路.本文研究基于距离-多普勒(Range?Doppler, RD)图的瞬态干扰自动识别方法,将RD图转化为灰度图,提取其纹理特征,再基于机器学习设计分类算法.首先,提出新的瞬态干扰模型,仿真产生干扰数据,避免训练依赖实测数据;其次,建立RD灰度图图库,分强干扰、弱干扰和无干扰三类情况;然后,提取局部二值模式(Local Binary Pattern, LBP)纹理特征,基于支持向量机设计二分类器,结合纠错输出编码设计三分类器.最后,通过实测数据和文献RD图,验证本文所提识别方法的准确性,并比较分析不同图像特征及参数的影响.

关键词: 超视距雷达, 距离?多普勒图, 瞬态干扰, 干扰识别, 支持向量机, 纹理特征

Abstract:

To improve the robustness of transient interference processing, over?the?horizon radar (OTHR) can take the way of suppression after assured detection. This paper analyzes automatic recognition of transient interference in the range?Doppler (RD) map, by transforming the RD map into gray image, extracting the texture features, and designing the classification algorithm based on machine learning. Firstly, a model of transient interference is developed to simulate the received data, so that the training does not rely on real data. Secondly, the image datasets are produced and classified into three categories, i.e. strong interference, weak interference, and non?interference. Then, the local binary pattern (LBP) texture features are extracted to design the binary classifier based on support vector machine (SVM) and then design the ternary classifier by error?correcting output codes (ECOC). Finally, simulations based on real data from OTHR and literatures demonstrate the effectiveness of our method and the effects of various parameters and image features.

Key words: OTHR, RD map, transient interference, interference recognition, SVM, texture features

中图分类号: