Face recognition based on 2D locally discriminating projection method

Locally discriminating projection (LDP) is a new subspace feature extraction method which takes special consideration of both the local information and the class information. For LDP method, the image matrix data are vectorized to find the intrinsic manifold structure, and the dimension of matrix data is usually very high, so LDP cannot be performed because of the singularity of scatter matrix. In addition, the matrix-to-vector transform procedure may cause the loss of some useful structural information embedding in the original images. Thus, in this paper, a novel method, called 2D locally discriminating projection (2DLDP), for face recognition is proposed. Experiments conducted on the ORL and FERET face database demonstrate the effectiveness of the proposed method.