主成分分析的模糊神经网络目标识别系统  

A Study on Target Identification based on Principal Component Analysis and Fuzzy Neural Network

作　　者：陈登峰[1,2] 嵇启春[1,2] 张洪才[1] 

机构地区：[1]西北工业大学自动化学院,陕西西安710072 [2]西安建筑科技大学信息与控制工程学院,陕西西安710055

出　　处：《火力与指挥控制》2008年第10期24-26,33,共4页Fire Control & Command Control

摘　　要：提出了一种基于模糊神经网络(Fuzzy Neural Networks,FNN)的目标识别算法。在对目标进行模糊化处理基础上,通过主成分分析(Principal Component Analysis,PCA)提取相应的特征空间,以畸变的特征向量对系统进行训练,从而获得较高的目标变化适应性。仿真实验结果表明,这种方法具有较强的自适应能力,抗噪性能也有所提高。This paper present a effective method for target identification based on Fuzzy Neural Network （FNN）. Firstly, the target images are processed for fuzzification and distortion. Secondly, the eigenspace is obtained based on Principal Component Analysis （PCA）. Lastly, the networks is trained with the distorted eigenvectors for higher adaptability of target variation. The results of simulation show that the correctness of this method is higher than that of networks with conventional training method. Obviously,the ability to counteract disturbance and noise is also raised.

关 键 词：目标识别 模糊神经网络 主成分分析 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]