Identification of Human Faces Using Orthogonal Locality Preserving Projections

This paper presents a technique for identification of human faces using an algorithm based on Orthogonal Locality Preserving Projections (OLPP). It differs from Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA), which preserves the Euclidean structure of face space. Locality Preserving Projections (LPP) finds an embedding that preserves local information, and obtains a face subspace that best detects the essential manifold structure. Locality Preserving Projections (LPP) is non-orthogonal, and this makes it difficult to reconstruct the data. This problem is overcome by using Orthogonal Locality Preserving Projection method which produces orthogonal basis functions and can have more locality preserving power than LPP. Since the locality preserving power is potentially related to the discriminating power, the OLPP is expected to have more discriminating power than LPP. This approach, builds an adjacency graph which best reflects the geometry of the face manifold and the class relationship between various points. The projection is then obtained by preserving such graph structure which forms the Orthogonal Laplacianface. In this way, the unwanted variations resulting from changes in lighting, facial expression and poses are reduced.