基于序列子空间的高分辨距离像噪声稳健识别方法

发布时间：2018-03-16 17:02

本文选题：雷达自动目标识别　切入点：高分辨距离像　出处：《中国科技论文》2017年02期 　论文类型：期刊论文

【摘要】：高分辨距离像(high resolution range profile,HRRP)是目标沿雷达视线方向上的一维压缩投影,传统的HRRP目标识别方法大都利用单次HRRP测试样本判决。但是,由于单次测试样本包含的信息有限,且容易受到噪声污染,识别鲁棒性难以保证。提出1种基于子空间的HRRP序列噪声稳健识别算法。该算法在训练、测试阶段均利用HRRP序列,基于主成分分析(principal component analysis,PCA)方法生成能够抑制噪声、冗余分量的目标信号子空间,并根据Grassmann流形定义子空间距离,将测试子空间与训练子空间按照最小子空间距离的准则作匹配比较,从而判定测试样本序列所属类别。文章推导证明了传统的最小重构误差方法是提出方法只使用单次HRRP测试样本的特殊情况。基于实测数据的识别实验显示,由于更充分地利用了HRRP序列信息且子空间能够抑制噪声,提出方法较最小重构误差方法具有更好的识别性能和噪声稳健性。
[Abstract]:High resolution resolution range profile is one dimensional compression projection of the target along the radar line of sight. Most of the traditional HRRP target recognition methods use single HRRP test samples to judge. However, because of the limited information contained in the single test sample, the high resolution range profile is one dimensional compression projection along the radar line of sight. It is easy to be polluted by noise, and the robustness of recognition is difficult to guarantee. A robust recognition algorithm for HRRP sequences based on subspace is proposed. The algorithm uses HRRP sequences in training and testing stages. Based on principal component analysis (PCA) principal component analysis (PCA) method, the target signal subspace which can suppress noise and redundant components is generated, and the subspace distance is defined according to the Grassmann manifold. The test subspace is compared with the training subspace according to the criterion of minimum space distance. In this paper, it is proved that the traditional method of minimum reconstruction error is a special case in which only a single HRRP test sample is used. Because the HRRP sequence information is fully utilized and the subspace can suppress the noise, the proposed method has better recognition performance and noise robustness than the minimum reconstruction error method.
【作者单位】：西安电子科技大学雷达信号处理国家重点实验室;西安电子科技大学信息感知技术协同创新中心;
【基金】：国家自然科学基金资助项目(61271024,61322103) 高等学校博士学科点专项科研基金资助项目(20130203110013) 陕西省自然科学基础研究计划(2015JZ016)
【分类号】：TN957.52

【相似文献】