A Novel Discriminant Analysis Approach Using Angular Fourier Transform for Face Recognition

In this paper, a novel discriminant analysis approach using Angular Fourier transform is proposed for face recognition. As a generalization of Fourier transform, the Angular Fourier transform is an important frequency-domain analysis technique. The proposed approach combines it with discriminant analysis method. First, this approach selects appropriate value of angle parameter for discrete Angular Fourier transform by using 2D separability judgment, and then it uses an improved Fisherface method to extract discriminative features from the preprocessed images. Finally, the nearest neighbor classifier is employed for classification. Using a public face databases as the test data, the experimental results demonstrate that the proposed approach outperforms several related discrimination methods.