基于支持向量域描述的雷达地面目标鉴别技术  

Radar Ground Target Discrimination Based on Improved SVDD

作　　者：李龙 LI Long(No. 20 Research Institute of CETC, Xi'an 710068)

机构地区：[1]西安导航技术研究所,西安710068

出　　处：《火控雷达技术》2021年第1期15-19,25,共6页Fire Control Radar Technology

摘　　要：在高分辨一维距离像目标识别中,有效的对库内目标特征空间进行描述,并且对库外目标进行鉴别是一个关键问题。本文提出了一种基于非均匀特征向量分布的目标鉴别器设计方法,该方法利用训练特征空间的协方差分布情况,选择库内协方差较小,即样本密度较大的区域进行细致描述,有效克服了训练样本分布非均匀造成的特征空间描述偏差,进而保证对库内目标的有限判断与库外目标的有效剔除,从而提升目标识别系统的总体性能,最后利用仿真与实测数据相结合的方式对该方法的性能进行了验证。Effective description of target feature space within the training database and recognition of targets outside the training database are key issues in high resolution range profile(HRRP)based target identification.In this paper,a target discriminator based on non-uniform feature vector distribution is proposed.The proposed target discriminator utilizes the covariance distribution of the training feature space to achieve more accurate feature space description.To be specific,the region with small covariance(i.e.,region with large sample density)in the training database is described in a more detailed way.Therefore,the proposed target discriminator can distinguish between targets within and without the training database effectively,improving the overall performance of the target identification system.At last,the simulation and test results prove that the proposed discriminator has good performance.

关 键 词：雷达 高分辨一维距离像 目标识别 支持向量域描述 

分 类 号：TN95[电子电信—信号与信息处理]