航空雷达目标极化分集特征的实时提取与自动目标识别的ANN技术  

Real-Time Polarizaton-Diverse Features Extraction and Automated Target Identification Using Neural Networks

作　　者：张良杰[1] 汪文秉[1] 

机构地区：[1]清华大学自动化系,西安交通大学信控系

出　　处：《电子学报》1994年第3期113-116,119,共5页Acta Electronica Sinica

摘　　要：本文首先利用Ｈｏｐｆｉｅｌｄ线性规划神经网络实现了一种极化椭圆匹配技术，从极化分集的雷达后向散射测量结果中提取了目标主散射体的全极化信息──极化分集特征；然后将实时获取的极化分集特征进行特征变换，进而直接利用基于欧氏测度的ＡＲＴ２神经网络进行自动目标识别．This paper presents a method that maps the polarization-diverse features extraction problem onto the Liapunov energy function of the Hopfield linear programming neural network,so as to obtain a real-time solution of these features derived from the polarization ellipse corresponding to the major scattering centers.Then those features are transformed so that we can make use of the ART2 neural network freely to perform automated target identification.

关 键 词：神经网络 特征 目标识别 航空雷达 

分 类 号：TN957.73[电子电信—信号与信息处理]