基于字典学习的人脸识别算法  

Face recognition algorithm based on dictionary learning

作　　者：任璐[1] 李祖松[1] Ren Lu;Li Zusong(West Anhui University,Lu’an Anhui,237012)

机构地区：[1]皖西学院,安徽六安237012

出　　处：《电子测试》2021年第9期63-64,132,共3页Electronic Test

基　　金：皖西学院校级自然科学研究项目(WXZR201625,WXZR202018)。

摘　　要：通常将能够表征人脸图像特征的特征集称为字典,进而利用字典可以获得图像的稀疏表示。目前,字典主要分为两类,一种是隐性字典,例如wavelet等;另一种是通过机器学习从样本中获取字典,可通过一种显性矩阵表示,例如K-SVD等。本文主要研究了K-SVD(Kernel-Singular Value Decomposition)字典学习算法,该算法适用性强,常与OMP算法结合使用。实验结果显示,该算法具有很好的收敛性以及较高的正确率。The feature set that can represent the features of face image is usually called a dictionary,and then the sparse representation of the image can be obtained by using the dictionary.At present,there are two kinds of representation:implicit dictionary and SVD dictionary.This paper mainly studies the K-SVD(kernel singular value decomposition)dictionary learning algorithm,which has strong applicability and is often combined with OMP algorithm.Experimental results show that the algorithm has good convergence and high accuracy.

关 键 词：人脸识别 稀疏表示 K-SVD OMP 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]