Skip to Main Content
Two-dimensional principal component analysis (2DPCA) for face recognition has been proposed which is based on 2D matrices. It needs more coefficients for feature vectors than principal component analysis (PCA). In this paper, we develop an idea which is working in the projective feature image obtained by 2DPCA on the original images i.e., image PCA, for efficient face representation and recognition. To test image PCA and evaluate its performance, a number of experiments are performed on two face image database: ORL and Yale face databases. The experimental results show that image PCA achieves the same or even higher recognition rate than 2DPCA, while the former needs less coefficients for feature vectors than the latter.