Palmprint recognition via Locality Preserving Projections and extreme learning machine neural network

Abstract

This paper proposes an efficient palmprint recognition method using locality preserving projections (LPP) and extreme learning machine (ELM) neural network. Firstly, two-dimensional discrete wavelet transformation (DWT) is applied in the region of interest (ROI) of each palmprint image and then principal component analysis (PCA) and LPP are used for dimensionality reduction. Finally, we construct a single-hidden layer forward network (SLFN) to construct one extreme learning machine (ELM) to quickly classify the palmprint images. Experiments on the PolyU palmprint database demonstrate the effectiveness of the proposed method.