摘要
文章综合传统的人脸识别方法,提出新的人脸识别算法。首先,建立近红外三维模型,采用傅里叶变换获取人脸三维信息。同时,建立可见光三维模型,采用3DMM技术获取三维人脸图像,然后将两种模型分别与参考系进行对比,最后进行融合,以此进行人脸识别。
Based on the traditional face recognition method, a new face recognition algorithm is proposed in this paper. Firstly, the nearinfrared 3D model is built and the face 3D information is acquired by Fourier transform. At the same time, the 3D model of visible light is established, and the 3D face image is obtained by using 3DMM technology. Then, the two models are compared with the reference system respectively, and finally fused for face recognition.
引文
[1]谢志华,伍世虔,方志军.LBP与鉴别模式结合的热红外人脸识别[J].中国图象图形学报,2012(6):707-711.
[2]曹林,李猛,张华.结合改进Surf特征的素描人脸识别[J].计算机工程与应用,2018(21):66-68.
[3]周先春,孙文荣,张纯伟.基于自适应对数变换和PCA算法的人脸识别研究[J].电子技术应用,2014(6):126-129.
[4]孙海江,王延杰,刘伟宁.基于自适应平台阈值和拉普拉斯变换的红外图像增强[J].中国光学,2011(5):474-479.
[5]CHEN X,FLYNN P J,BOWYER K W.Visible-light and infrared face recognition[J].In Proceedings IEEE Workshop on Multimodal User Authentication,2003(12):48-55.
[6]李杰.可见光/近红外人脸识别方法研究与实现[D].北京:北京交通大学,2016.
[7]潘磊,尹义龙,李徐周.基于得分的近红外线与可见光图像融合算法[J].计算机工程,2013(4):226-230.