Face feature extraction and recognition using contourlet transform and coupled subspace analysis

Contourlet transform has good properties of energy aggregation, multiresolution and directional image expansion. Coupled Subspace Analysis (CSA) utilizes optimal bi-directional projection based matrix to deduce the dimension, which enables it to obtain better recognition result and lower computational complexity. In this paper, an efficient face feature extraction and recognition method using contourlet transform and CSA is presented. Firstly, each face is decomposed by contourlet transform. Then frequency coefficients in the same scale and various directions are fused into a subband. Finally face discriminant features are extracted in fused subbands by improved CSA. Experiments on ORL and PIE and show that our method is effective and can obtain reliable correct recognition rate.