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

作　　者：陈非[1] 黄山[1,2] 张洪斌[2] 

机构地区：[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

出　　处：《计算机应用与软件》2015年第4期156-159,共4页Computer Applications and Software

摘　　要：针对传统的人脸识别算法在每个人只有单个训练样本时识别性能严重下降的问题,提出了通用学习框架改进核主成分分析的单样本人脸识别算法。首先,选取一个合适的通用训练样本集,将各个单训练样本与通用训练样本集中某人的多训练样本按比例叠加;然后,利用经典的KPCA算法进行特征提取,将所有叠加后的训练样本和测试样本投影到特征子空间;最后,使用最近邻分类器完成最终的人脸识别。在Yale及FERET两大通用人脸数据库上的实验结果表明,相比其他几种较为先进的人脸识别算法,该算法取得了更好的单样本识别效果。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 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]