A New Separability Criterion for Face Recognition Based on Wavelet Packet

Face recognition based on wavelet packet usually selects the best bases with the criterion of the difference of class energy, and this criterion is not the best one in pattern classification. So this paper proposes a new separability criterion in order to the extracted features which can be classified better. And then we also combine the LDA method so as to suppress the difference within class while keeping the difference between classes. Experiments show that the method can achieve higher recognition rate.