Measuring biometric feature information in palmprint images

Abstract

The measurement of biometric feature information is important for biometric technology, as for it can determine the uniqueness of biometric features, compare the performance of several feature extraction methods and quantify whether combination of features or biometric fusion offers any advantage. In this paper, we study the measurement of palmprint feature information using relative entropy between intra-person and inter-population. We compute the biometric feature information in which the feature extracted by three different methods, including: Principal Component Analysis (PCA), Linear Discriminant Analysis(LDA) and Locality Preserving Projections(LPP). The average biometric feature information is calculated to be approximately 280 bits for PCA, 246 bits for LDA features and 460 bits for LPP. © Springer International Publishing 2013.