基于深度学习的HRRP舰船目标识别性能分析  

Performance Analysis of HRRP Ship Target Recognition Based on Deep Learning

作　　者：黄柏圣[1] 江涛[1] 周伟光[1] HUANG Baisheng;JIANG Tao;ZHOU Weiguang(Nanjing Research Institute of Electronics Technology,Nanjing 210039,China)

机构地区：[1]南京电子技术研究所,南京210039

出　　处：《现代雷达》2020年第2期54-57,共4页Modern Radar

摘　　要：针对分辨率、信噪比、补零及深度置信网络等对深度学习舰船目标识别性能的影响问题,文中开展了基于实测数据的相关实验分析,整个实验分析处理过程包括回波信号对齐、数据脉冲压缩、信号能量归一化、深度学习模型训练、分类器设计及判决输出。实验分析结论为深刻理解基于深度学习的高分辨距离像舰船目标识别技术原理内涵,开展舰船目标识别工程化应用设计奠定了坚实的基础。Aiming at the influence of resolution,signal to noise ratio,null and deep belief network on the performance of deep learning ship target recognition,the performance analysis of correlation recognition based on measured data is carried out. The whole experimental analysis and processing includes echo signal alignment,data selling pulse pressure processing,signal energy nor-malization,depth model training,classifier design and decision output. The conclusion of the experiment has laid a solid foundation for deep understanding of the principle of deep learning high resolution range profile ship target recognition technology and the engineering application design of ship target recognition.

关 键 词：高分辨率距离像 深度学习 目标识别 分类器 

分 类 号：TN957.51[电子电信—信号与信息处理]