基于二维主成分分析法的人脸识别技术性能分析

Performance Analysis of Face Recognition Technology Based on Two-dimensional Principal Component Analysis

人脸识别系统由人脸图像预处理、人脸图像特征提取以及匹配识别和人脸图像采集检测4个部分组成。为剖析二维主成分分析法在不同维度和样本数下的识别效果,对ORL数据库和Yale数据库的人脸数据样本进行训练与识别。将二维主成分分析法与主成分分析法、线性判别分析法和独立成分分析法等3种人脸识别算法进行比较。结果表明,样本数与2DPCA算法的识别率均呈正相关关系。相同样本数下,同样特征维数间隔时,按平均识别率由大至小排列为:2DPCA、ICA、LDA、PCA,其中2DPCA识别率最高,为0.62,PCA识别率最低,为0.45。

The face recognition system consists of four parts:face image preprocessing,face image feature extraction,matching recognition and face image acquisition detection.In order to analyze the recognition effect of two-dimensional principal component analysis(2DPCA)in different dimensions and sample numbers,face data samples from ORL and Yale databases are trained and recognized.Two-dimensional principal component analysis is compared with principal component analysis(PCA),linear discriminant analysis(LDA)and independent component analysis(ICA).The results show that the number of samples is positively correlated with the recognition rate of 2DPCA algorithm.With the same number of samples and the same feature dimension interval,the average recognition rates for each algorithm are arranged from large to small as follows:2DPCA,ICA,LDA,PCA,of which the highest value is 0.62 and the lowest is 0.45.