Two-dimensional discriminant locality preserving projections for face recognition

One of the key issues of face recognition is to extract the features of face images. 2D-DLPP is a new feature extraction method for face recognition. 2D-DLPP benefits from three techniques, i.e., locality preserving projections (LPP), image based projection and discriminant analysis. Firstly, LPP can optimally preserve the local structure of the samples. Secondly, compared to vector based projection, image based projection can avoid the small sample size problem and give more spatial structural information of image. Finally, discriminant analysis applied in 2D-DLPP can improve recognition performance by maximizing the interpersonal distance and minimizing the intrapersonal distance. The experimental results show that 2D-DLPP is robust and has better face recognition performance than other methods. In addition, row projection and column projection of 2D-DLPP are discussed and compared. Though the expressions of the two projections are similar, the experimental results of them are different. Therefore, it is necessary to select a suitable projection method for a certain face image database.