基于LDA的眉毛识别方法研究  

Research on Eyebrows Recognition Based on Linear Discriminant Analysis

作　　者：李颜瑞[1] LI Yanrui(Information Engineering Department,Shanxi Institute of Mechanical and Electrical Engineering,Changzhi 046011,China)

机构地区：[1]山西机电职业技术学院,信息工程系,山西长治046011

出　　处：《微型电脑应用》2022年第11期38-40,44,共4页Microcomputer Applications

基　　金：山西省高等学校科技创新项目(2020L0755)。

摘　　要：基于LAD提出了一种眉毛识别方法。首先利用线性判别分析方法进行特征提取,是在主成分分析方法(PCA方法)提取特征的基础上,再进行两次投影得到最终的特征向量。解决了PCA中忽略高阶统计量信息的缺点。最后利用最近邻方法进行识别。在109人的眉毛数据库上进行实验,得到了比较满意的识别率,进一步证明了眉毛识别的可行性和正确性。This paper presents an eyebrow recognition method based on linear discriminant analysis.Firstly,linear discriminant analysis is used to extract features,and based on independent component analysis,features are re-extracted.The final eyebrow feature is obtained by the two projections.The method solves the problem of ignoring the information of higher-order statistics in independent component analysis.Finally,the nearest neighbor method is used for recognition.The experiment results show that in an eyebrow information database with 109 persons,this method obtains a satisfactory recognition rate,which further proves the feasibility and correctness of eyebrow recognition.

关 键 词：眉毛识别 线性判别分析 特征提取 最邻近距离法则 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]