论文部分内容阅读
针对线性判别分析(LDA)的“小样本”和要求数据须服从高斯分布的问题,提出一种基于非参数化最大间隔准则(NMMC)的雷达目标识别方法.首先,利用自相关小波变换提取目标高分辨距离像(HRRP)的非平稳特征,将其与HRRP原信号一起作为目标的分类特征,利用NMMC实现特征提取;然后,通过支持向量机进行分类.NMMC在解决小样本问题的同时,松弛了对数据分布的类高斯要求.最后,基于5种飞机高分辨距离像数据的仿真实验验证了所提出方法的有效性.
Aiming at the problem that the “small sample” of linear discriminant analysis (LDA) and the data to be obeyed by Gaussian distribution, a new radar target recognition method based on non-parametric maximum separation criterion (NMMC) is proposed.Firstly, by using autocorrelation wavelet transform The non-stationary features of target HRRP are extracted and used as target classification features together with HRRP original signals, and the features are extracted by using NMMC.Secondly, SVM is classified by support vector machine (SVM) , Relaxed the Gaussian requirement for data distribution.Finally, the simulation experiments based on the five kinds of aircraft high-resolution range image data verify the effectiveness of the proposed method.