局部二维主元分析的人脸识别新方法

A New Face Recognition Approach:Local Two-dimensional Principal Component Analysis

人脸姿态、表情、光照等变化是影响人脸识别的主要因素,如何减轻这些因素对识别率的影响是人脸识别的研究关键所在。R.G等人提出了MPCA方法,通过对人脸图像进行一次分块处理,减少了这些因素产生的影响。然而MPCA方法只消除了部分影响,仍未能完全解决这一问题。文章提出了一种进行二次分块处理的新方法——局部二维主元分析方法,进一步消除了这些因素所产生的影响。通过在Yale国际标准人脸库及UMIST人脸库上进行验证,该方法大大提高了人脸识别率。

Varieties of pose ,expression and illumination are the factors which impact the face recognition all the time. MPCA method,which could reduce these impacts by dividing the face image into blocks once,was proposed by R.G.In order to reducing these impacts further and a new method is proposed.The experiment results indicate that the novel method is effective.