Two-dimensional nearest neighbor discriminant analysis

Recently, some feature extraction methods have been developed by representing images with matrix directly, however few of them are proposed to improve accuracy of classification directly. In this paper, a novel feature extraction method, two-dimensional nearest neighbor discriminant analysis (2DNNDA), is proposed from the view of the nearest neighbor classification, which makes use of the matrix representation of images. We apply 2DNNDA to face recognition and the results demonstrate that 2DNNDA outperforms the conventional methods.