计算机应用与软件Issue(4)：156-159,4.DOI:10.3969/j.issn.1000-386x.2015.04.038
通用学习框架改进核PCA的单样本人脸识别
SINGLE SAMPLE FACE RECOGNITION WITH KERNEL PCA OPTIMISED BY GENERIC LEARNING FRAMEWORK
陈非 1黄山 1张洪斌2
作者信息
- 1. 四川大学电气信息学院 四川 成都610065
- 2. 四川大学计算机学院 四川 成都610065
- 折叠
摘要
Abstract
For the problem that the recognition performance of traditional face recognition algorithms degrades seriously when each person has only one training sample,we propose a single sample face recognition algorithm which uses generic learning framework to improve kernel principle component analysis (KPCA).First,it selects a suitable generic training sample set and superposes each single training sample over the multiple training sample of a certain person in generic training set in proportion.Then,it uses typical KPCA to extract the features and projects all superposed training and testing samples onto feature subspace.At last,it uses the nearest neighbour classifier to complete the finale face recognition.Experiments results on two popular face databases of Yale and FERET show that the proposed algorithm achieves better recognition effect on single sample than several other relatively advanced face recognition algorithms.关键词
人脸识别/单样本每人/通用学习框架/最近邻分类器/核主成分分析Key words
Face recognition/Single sample per person/Generic learning framework/Nearest neighbour classifier/Kernel principle component analysis分类
信息技术与安全科学引用本文复制引用
陈非,黄山,张洪斌..通用学习框架改进核PCA的单样本人脸识别[J].计算机应用与软件,2015,(4):156-159,4.
