Principal component analysis of gender, ethnicity, age and identity of face images

Abstract

Abstract—Principal Component Analysis (PCA) has been widely used for efficient representation of face images data in a low dimensional subspace. In this study, we use PCA to analyse different properties of faces, such as gender, ethnicity, age and identity. Using Linear Discriminant Analysis (LDA), we show that PCA efficiently encodes information related to different properties, different components of PCA encode different information, and there may be components which encode information related to multiple properties. I.