融合均匀局部二元模式和稀疏表示的人脸识别  

Face recognition integrated ULBP and sparse representation

作　　者：高洪涛[1] 郜亚丽[2] GAO Hong-tao;GAO Ya-li(Department of Cyber Crime Investigation,Criminal Investigation Police University of China,Shenyang 110035,China;Adult Education Center,Jiyuan Vocational and Technical College,Jiyuan 459000,China)

机构地区：[1]中国刑事警察学院网络犯罪侦查系,辽宁沈阳110035 [2]济源职业技术学院成教中心,河南济源459000

出　　处：《计算机工程与设计》2019年第3期869-873,共5页Computer Engineering and Design

基　　金：国家科技支撑计划基金项目(2007BAK34B03)

摘　　要：针对当前人脸识别受噪声等干扰导致识别效果不佳的问题,提出一种人脸识别方法,利用具有抗光照不变性的近红外人脸图像进行人脸识别算法分析。为增强算法在图像抗噪声方面的性能,通过均匀局部二元模式提取人脸图像的特征;为避免细节纹理的缺失,采用对人脸图像分块处理的方法建立特征;通过引入稀疏分类算法,进行人脸识别。实验结果表明,该人脸识别算法相比传统方法具有良好的稳健性与识别率,达到了提高人脸识别算法性能的目的。In view of the poor results of the current face recognition caused by noise and other problems, a face recognition method was presented. The near-infrared face image with anti-illumination-invariant for the research of face recognition algorithm was used. To enhance the performance of anti-noise, the uniform local binary pattern was utilized to extract face features. The face image was blocked to avoid the detail texture missing and construct face features. A method of sparse representation for face classification was used. Experimental results demonstrate that the proposed method is robust with better recognition compared with the traditional method. The face recognition algorithm can benefit a lot from the proposed approach.

关 键 词：特征提取 人脸识别 近红外人脸 稀疏 均匀局部二元模式 

分 类 号：TP393.03[自动化与计算机技术—计算机应用技术]