基于多模态平稳序列建模的雷达高分辨距离像有限样本目标识别方法

A New Method Based on Multimodal Stationary Sequence Modeling for Radar HRRP Target Recognition under Small Training Set Conditions

为了解决雷达高分辨距离像识别系统对训练样本需求量过大的问题,提出了一种有限样本条件下的目标识别新方法。分析了距离像频谱幅度的统计特性,从其广义平稳性和多模态分布特性出发,定义一种线性混合高斯状态空间模型对其统计建模,利用期望最大化算法进行了模型参数估计。实验结果表明:即使在很少的训练样本条件下,该方法仍能获得较高的正确识别率和良好的拒判性能。

A new recognition method is developed to relax the heavy requirement of training sample size in the radar high resolution range profile（HRRP） target recognition.The statistical characteristics of HRRP′s frequency spectrum amplitude（FSA） are analyzed.Then a linear Gaussian mixture state space model（LGMSSM） is proposed to describe the stationarity and multimodality of the FSA.Afterwards,an expectation maximization（EM） algorithm is derived for model parameter estimation.Finally,experimental results based on measured data show that the proposed method can obtain satisfactory recognition accuracy and rejection performance even with a very few training samples.