International Journal of Computer Theory and Engineering

Abstract

Locality Preserving Projection (LPP) aims to preserve the local structure of the image space, while Principal Component Analysis (PCA) aims to preserve the global structure of the image space; LPP is linear, while Isomap, LLE, and Laplacian Eigenmap are nonlinear methods, so they yield maps that are defined only on the training data points and how to evaluate the maps on novel test data points remains unclear. Locally Discriminating Projection (LDP) is the extension of LPP, which seeks to preserve the intrinsic geometry structure by learning a locality preserving sub manifold. LDP is a new subspace feature extraction method and supervised because it considers both class and label information. LDP performs much better than the other feature extraction methods such as PCA and Laplacian faces. In this paper an extension to LDP called Wavelet based Kernel Locally Discrimination Projection (-WKLDP) is proposed to extract non linear features of sub band face images for classification, where as LDP considers linear features only. In the proposed method first by using wavelets the sub band face images are constructed, then on sub band face images kernel locally discriminating projection (KLDP) is applied. The experimental results on the ORL face database suggest that W-KLDP gives lower time complexity and have high recognition rates than other existing methods.