一种基于指纹人脸的多生物特征身份认证方法

摘要

This paper introduces a kind of multimodal biometrics recognition approach based on fingerprint and human face.In addition,we come up with an improved algorithm based on the traditional fingerprint and human face method.A method is put forward which combines local normalization with gaussian filter.This method is used to calculate the direction of the fingerprint,and then calculate by using the fast local ridge compensation method which can identify the fingerprint more quickly and more accurately.Human Face recognition makes a matching by locating the human＇s faces first and then extracting the feature of human face.This paper uses LBP（Local binary patterns） operator to extract the face features and to reduce the dimension of the human face by PCA in advance.Also we adopt the extreme learning machine to make a matching.By combining the results of the identification of human face and fingerprint for judging,a correct and steady identification system is achieved.%本文分别介绍了一种基于指纹人脸识别的多生物特征身份认证方法,并针对传统的指纹人脸方法提出相应的改进算法。对指纹识别,本文提出采用局部归一化方法结合Gaussian滤波器来计算指纹方向,再对局部脊线补偿法（Loca lRidge Compensation）进行快速运算,能够更加快速准确地进行指纹识别。人脸识别通过定位人脸位置并且进一步提取人脸特征来进行匹配,使用LBP（Local binary patterns）算子对人脸样本进行局部特征提取,对LBP处理后的人脸图像使用主成分分析（PCA）进行降维,并采取了极限学习机（Extreme learning machine,ELM）分类器进行匹配,将指纹、人脸的识别结果在决策层进行融合,最后做出判断,从而得到准确稳定的身份认证系统。