摘要
实时性和识别率是评估SAR图像目标识别系统性能的两个主要指标。分析了影响这两个指标的关键因素,并以此为基础,提出了一种快速的SAR图像目标识别方法。该方法采用基于Hebb学习规则的自组织神经网络提取主分量特征,使用多层感知器神经网络(MLP NN)进行目标分类。实验对比分析表明,在识别率较高的同时,该方法具有内存需求少、运行速度快的特点,能用于实时处理。
The real-time ability and recognition rate are two primary goals for evaluating the performance of an SAR imagery target recognition system. The key factors which influence these two goals are analyzed. According to the analysis, a fast SAR target recognition approach is proposed, which utilizes an auto-organized neural network to extract the principal component features and a multi-layer neural perceptron network （MLP NN） as the classifier. The experimental results show that it consumes relatively less memory and runs very fast with a considerable recognition rate, thus can be used in the real-time situation.
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2008年第10期1855-1859,1943,共6页
Systems Engineering and Electronics
基金
国家自然科学基金资助课题(60772045)
