Abstract: Though principal component analysis (PCA) is a classical method for face recognition,the PCA method extracts global features of the original images,and it does not consider the local discriminant features.In contrast,Modular PCA method extracts the important local discriminant features,and it achieves better performance than the PCA method in face recognition.However,vectorization in PCA or modular PCA often causes "curse of dimensionality".In order to extract features from matrix or higher-order tensor objects directly,multilinear principal component analysis (Multilinear PCA) is developed.Multilinear PCA can avoid "curse of dimensionality",meanwhile it will not destroy the original data structure.Inspired by Modular PCA and Multilinear PCA,we proposed a new method called modular multilinear principal component analysis (M²PCA) for face recognition.Experiments were conducted on the Yale,XM2VTS and JAFFE databases respectively,and experimental results indicate that,under the same condition of sub-blocks,the proposed method is obviously superior to the general Modular PCA.

Key words: Face recognition,Feature extraction,Multilinear PCA,Modular PCA