Combining DCT and LBP Feature Sets For Efficient Face Recognition

In this paper, we present a novel approach for face recognition combining classifiers based on both micro texture in spatial domain provided by local binary pattern (LBP) and macro information in frequency domain acquired from the discrete cosine transform (DCT) to represent facial image. The classification of these two feature sets is performed by using support vector machines (SVMs), which had been shown to be superior to traditional pattern classifiers. The experiments clearly show the superiority of the proposed classifier combination approaches over individual classifiers on the Yale face database and a high correct classification rate of 96% is obtained.