基于BP神经网络的指纹模板分类器分类算法  

A New Classifiable Algorithm of Fingerprint Template Based on Backpropagation Neural Networks

作　　者：蔡俊[1] 任德官[1] 

机构地区：[1]浙江师范大学计算机科学研究所,金华321004

出　　处：《微电子学与计算机》2002年第9期1-3,共3页Microelectronics & Computer

基　　金：浙江省自然科学基金资助项目(ZD0108)

摘　　要：指纹识别是计算机模式识别领域中一个比较活跃的课题,有着十分广泛的应用前景。对于庞大的指纹信息,人们越来越关心的是如何对其分类与储存。文章提出了一种基于BP神经网络的指纹模板分类算法,在简介指纹图像的预处理和模板建立过程的基础上,着重阐述基于黄金分割法的自适应变步长算法。仿真表明它比传统的固定步长算法有更好的收敛速度与精度。Fingerprint recognitio n is an active subject in the field of pattern recognition which has an extensiv e applied foreground.People are more and more concerned about how to classify a nd how to store the huge number of fingerprint template information.The paper p resents a classifiable algorithm of fingerprint template based on BP neural netw orks.Having simply introducing the process of image precondition and fingerprin t template building,we emphatically formulate a self-adapting and step-changi ng algorithm used gold-segmentation.At last,the simulation result indicates t hat the new algorithm is more superior in convergent rate and precision than tha t of other traditional algorithms.

关 键 词：BP神经网络 指纹模板分类器 分类算法 指纹识别 模式识别 特征提取 计算机 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] D918.91[自动化与计算机技术—计算机科学与技术]