Experimental Evaluation of Different Intensity Distributions for Palmprint Identification

Abstract

The purpose of this paper is to investigate the influence of different intensity distributions on palmprint identification. A intensity adjustment function, which can overcome the shortage of intensity translation caused by unstable lighting, is used to generate intensity distributions. Experiments, which are based on the database of 98 individuals, using Gabor features and PCA features show that the performances of each experiment are varied more or less, and they perform better when the mapping weighted toward darker than the mapping weighted toward brighter. Two assumptions are considered that when the mapping weighted toward darker, palmprints have more distinction (or contrast) for individuals; and the intensity distributions have more consistency which can overcome the shortage brought by the unstable lighting. They would be validated in future work. That the Gabor features perform better than PCA features is an additional conclusion of this paper.