摘要
本文介绍了多态神经元模型和复数Hopfield神经网络,给出了网络权值确定的两种方法:复数Hebbian学习规则和内积法构造厄米矩阵。另外借助于二维离散傅立叶变换给出了灰度图相位变换和反变换的方法,实现了复数Hopfield神经网络对灰度图的存储。在此基础上借助于网络的联想记忆特性,将其应用于交通路牌的识别中。
A multistate neuron and complex valued network are introduced.Tow ways are given to determine its weight mastics:complex valued Hebbian rules and hermitian mastics inner product.By using the tool of 2-dimensional Fourier Transform,image phase transform and its inverse transform,the storage of gray scale image is realized.Take advantage of the associative property of the network,it is used in road sign recognition.
出处
《微计算机信息》
2010年第22期161-163,共3页
Control & Automation
关键词
复数Hopfield神经网络
多态神经元
灰度图
相位变换
路标识别
Complex-Valued Hopfield Neural Network(CVHNN)
Multistate Neuron
Gray Scale Image Storage
Image Phase Transform
Road Sign Recognition