Abstract: According to the target echo of low-resolution radar, a feature extraction method based on the theory of compressed sensing was proposed in this paper. An orthonormal wavelet basis was selected as the sparsity basis and a random sub-Gaussian matrix as the measurement matrix, and the feature extraction vector was composed of a few measurements. The proposed algorithm can not only get the intrinsic driving source of radar echo signal, but also keep the structure of the original signal and enough target information. The experimental results show that the feature extraction approach offers low dimension and high information density, and gets better result.

Key words: compressed sensing, feature extraction, low-resolution radar, target recognition

摘要： 根据低分辨雷达目标回波的特点，提出了一种基于压缩传感的回波特征提取方法。选择小波基为稀疏基，高斯随机矩阵为测量矩阵，由较少的测量值构成识别特征向量，不仅可以获得雷达回波信号的本质驱动源，而且能够保持原始雷达回波信号的结构和足够多的目标信息。实验结果表明，该算法提取的回波特征向量维数低，且信息密度高，可以得到较好的目标识别结果。

关键词: 压缩传感, 特征提取, 低分辨雷达, 目标识别