摘要
提出了利用径向基函数网络(RBF)识别人脸的方法,分析了其识别原理,讨论了网络结构的设计,给出了各层节点,隐居中心向量及形状参数的选取原则,并针对ORL人脸库的多幅人脸图像进行了识别实验。实验结果表明,RBF网络的识别效果较好,且其学习算法要比BP网络的学习算法简单。
In this paper, we applied Radial Base Functions Network to recognize human face, and analysed the recognition principle. The design of network structure, as well as the problem of choosing node numbers in every layer, node centre vectors and shape paramater in hidden layer are discussed in details. Expriments have been conducted based on face photoes in ORL face database. The expriments results show the recognition effect is fairish, and the study algorithm of RBF is simpler than the BP algorithm.
出处
《河南科学》
2003年第3期290-294,共5页
Henan Science
关键词
径向基函数网络
人脸识别
感受野
人脸库
开关参数
Radial Base Functions Network
face recognition
receptive field
face database
