Journal of Systems Engineering and Electronics ›› 2012, Vol. 34 ›› Issue (3): 489-494.doi: 10.3969/j.issn.1001-506X.2012.03.11

• 传感器与信号处理 • 上一篇    下一篇

基于ISVDD的雷达高分辨距离像在线识别方法

冯国瑜1,肖怀铁1,付强1,关卿2   

  1. 1. 国防科学技术大学电子科学与工程学院ATR实验室, 湖南 长沙 410073;
    2. 中国人民解放军63886部队, 河南 洛阳 471003  
  • 出版日期:2012-03-22 发布日期:2010-01-03

Online recognition method of HRRP for radar systems based on ISVDD

FENG Guo-yu1, XIAO Huai-tie1, FU Qiang1, GUAN Qing2   

  1. 1. ATR Lab, School of Electronic Science and Engineering, National University of Defense Technology, Changsha 410073, China;
    2. Unit 63886 of the PLA, Luoyang 471003, China
  • Online:2012-03-22 Published:2010-01-03

摘要:

现有高分辨距离像(high resolution range profile, HRRP)在线识别方法的训练过程往往需要大规模的训练样本集,难以在小规模样本条件下进行在线识别。针对小规模HRRP样本集的在线识别问题,推导了增量支持向量数据描述(incremental support vector data description, ISVDD)算法,并提出了基于ISVDD的HRRP在线识别方法。该方法在训练过程避免了对大规模样本集的需求,能够获得良好的识别效果。同时,由于ISVDD应用于在线识别,相对于标准支持向量数据描述在线识别方法,大大减少了新增样本的训练时间,而且能够获得和成批支持向量数据描述识别方法相同的识别性能。基于4种飞机目标的实验结果证明了本文方法的正确性和有效性。

Abstract:

The training process of high resolution range profile (HRRP) online recognition methods in existence usually needs large scale HRRPs, and these methods are difficult to achieve online recognition under small scale HRRPs condition. In order to deal with the online recognition issue with small scale HRRPs, an incremental support vector data description (ISVDD) algorithm is firstly developed, and then an online recognition method of HRRP for radar systems based on ISVDD is proposed. The proposed method can avoid the requirement of large scale samples in training process and achieves an excellent recognition performance. Meanwhile,since ISVDD is used to perform online recognition, compared with the online recognition method based on standard SVDD, the proposed method reduces the training time of incremental samples and can get the recognition performance as well as that of HRRP recognition in batches based on SVDD. The experimental results based on four types of planes prove the efficiency and validity of the proposed method.

中图分类号: