摘要
针对光照变化、姿态变化等条件下人脸鲁棒性差的难题,提出了一种改进主成分分析与低秩投影的鲁棒性人脸识别算法。首先利用改进主成分分析对人脸图像进行学习,形成低秩稀疏误差矩阵,然后根据稀疏误差图像计算平滑度和边缘,并进行加权实现人脸识别,最后进行仿真实验。结果表明,相对于当前经典人脸识别算法,本文算法获得更高的人脸识别率,并且具有较强的鲁棒性。
To solve the face poor robust problem in illumination changes,pose variations and other conditions,this paper proposes a robust face recognition algorithm based on improved principal component analysis and the low rank projection. First of all,the improved principal component analysis is used to learn the training samples and obtains low rank sparse error matrix,and then through the calculation of sparse error image smoothness and edge to achieve face recognition,finally the simulation experiments are carried out on Yale and AR face database. The experimental results show that,relative to the current classic face recognition algorithm,the proposed algorithm can not only improve the face recognition rate,and has strong robustness.
出处
《激光杂志》
北大核心
2015年第7期68-71,共4页
Laser Journal
基金
浙江省中青年学科带头人学术攀登项目(pd2013435)
浙江省教育厅一般(Y201430818)
关键词
主成分分析
人脸识别
低秩投影
特征提取
principal component analysis
face recognition
low-rank projection
features extraction
