摘要
介绍一种利用径向矢量提取形状特征的方法，着重于分析二维图形的不变性，提出一种带有方向因子的径向矢量描述．该矢量以图形边界弧长为自变量，完整地刻画了图形特点，克服了非凸图形识别中存在的多义性．以归一化的特征矢量为输入，采用三层神经网络为分类器。
A feature extracting method based on radius vectors was introduced.By using a directional factor,the proposed feature gives a complete description of the shapes,and it eliminates the ambiguity that may occur in the traditional shape representation of concave shapes.The normalized features can be used for final classification.With the help of a three layer BP neural network,satisfactory results in recognizing characters and military objects were carried out.
出处
《红外与毫米波学报》
SCIE
EI
CAS
CSCD
北大核心
1996年第4期250-256,共7页
Journal of Infrared and Millimeter Waves
基金
国家自然科学基金
关键词
形状分类
径向矢量
不变性识别
神经网络
shape classification,radius vector,invariant recognition,neural network.
