Face recognition using DT-CWT feature-based 2DIFDA

This paper introduces a novel DT-CWT feature-based Two-dimensional Inverse FDA (2DIFDA) by integrating the Dual-Tree Complex Wavelet Transform (DT-CWT) of face images and 2DIFDA method for face recognition. The DT-CWT has approximate shift invariance, good directional selectivity and can provide effective feature representation for face images. In the proposed method, DT-CWT is first used to extract the face image features at different scales and orientations. 2DIFDA is then applied for feature selection and dimensionality reduction in the DT-CWT feature space. Experimental results on ORL and FERET face databases demonstrate the feasibility of the new method.