A Spectral Domain Local Feature Extraction Algorithm for Face Recognition

In this paper, a spectral domain feature extraction algorithm for face recognition is proposed,which efficiently exploits the local spatial variations in a face image. For the purpose of featureextraction, instead of considering the entire face image, an entropy-based local band selectioncriterion is developed, which selects high-informative horizontal bands from the face image. Inorder to capture the local variations within these high-informative horizontal bands precisely, afeature selection algorithm based on two-dimensional discrete Fourier transform (2D-DFT) isproposed. Magnitudes corresponding to the dominant 2D-DFT coefficients are selected asfeatures and shown to provide high within-class compactness and high between-classseparability. A principal component analysis is performed to further reduce the dimensionality ofthe feature space. Extensive experimentations have been carried out upon standard facedatabases and the recognition performance is compared with some of the existing facerecognition schemes. It is found that the proposed method offers not only computational savingsbut also a very high degree of recognition accuracy.