基于高维特征域的低分辨雷达小微目标分类识别方法  

Classification and recognition method of small and micro targets in low resolution radar based on high dimensional feature domains

作　　者：徐好 吴琳拥 周云 任浩浩 XU Hao;WU Linyong;ZHOU Yun;REN Haohao(Sichuan Jiuzhou Falcon Technologies Co.,Ltd.,Mianyang 621000,China;School of Information and Communication Engineering,University of Electronic Science and Technology of China,Chengdu 611731,China)

机构地区：[1]四川九洲防控科技有限责任公司,绵阳621000 [2]电子科技大学信息与通信工程学院,成都611731

出　　处：《电子科技大学学报》2025年第2期203-209,共7页Journal of University of Electronic Science and Technology of China

基　　金：国家自然科学基金(42027805)。

摘　　要：低空小微目标分类问题是雷达业界的难题之一,严重影响了雷达的探测性能和系统作战指挥效能。为了准确、快速识别旋翼、固定翼等低空小微目标,提出一种基于高维特征域的低分辨雷达小微目标分类识别方法。通过提取信号层的一系列时频微观特征和航迹宏观特征,对特征进行内积、幂变换等获取高维特征域,利用学习树网络建立多层级目标分类识别模型,实现低空小微目标分类标记。研究结果表明,该方法能准确、快速地实现小微目标的分类。The classification of small and micro targets at low altitude is one of the difficult problems in radarfield,which seriously effects the detection performance of radar and the effectiveness of system combat command.In order to accurately and quickly identify small and micro targets at low altitude such as rotors,fixed wings andvehicles,a classification and recognition method of small and micro targets of low-resolution radar based on high-dimensional feature domain is proposed in this paper.A series of time-frequency micro features and track macrofeatures are extracted from the signal layer,and high-dimensional feature domain is obtained by internal productand power transformation of features.A multi-level target classification and recognition model is established byusing learning tree network to realize the classification and marking of small and micro targets at low altitude.Theresults show that this method can classify small and micro objects accurately and quickly.

关 键 词：小微目标 低分辨雷达 高维特征 分类识别 学习树网络 

分 类 号：TN957[电子电信—信号与信息处理] TN958[电子电信—信息与通信工程]