Weighted contourlet binary patterns and image-based fisher linear discriminant for face recognition

We propose a novel face representation model, called the weighted Contourlet binary patterns (WCBP), based on the NonSubsampled Contourlet Transform (NSCT), for face recognition. The decomposition using NSCT can capture rich image information at multiple scales, orientations, and frequency bands. This guarantees its robustness to illumination and expression variations. The weighting scheme embeds different discriminative powers of each NSCT-decomposed image. We also propose to carry out a subsequent Fisher linear Discriminant (FLD) on each decomposed image (named as WCBP+FLD) for dimension reduction of features. Our extensive experiments on the public FERET, CAS-PEAL-R1 and LFW databases demonstrate that the non-weighted Contourlet binary patterns performs better than local Gabor binary patterns. WCBP further improves the recognition rates. WCBP+FLD can achieve much competitive or even better recognition performance compared with the state-of-the-art Gabor feature based face recognition methods.