Nonlinear DCT Discriminant Feature Extraction with Generalized KDCV for Face Recognition

A nonlinear DCT discriminant feature extraction approach for face recognition is proposed. First, we analyze the nonlinear discriminabilities of DCT frequency bands and select appropriate bands. Second, we extract nonlinear discriminant features from the selected DCT bands by presenting a new kernel discriminant method, i.e. generalized kernel discriminative common vector (KDCV) method. The experimental results on the Feret database demonstrate the effectiveness of this new approach.