Noise-Robust Hand Based Multimodal Biometric System Using Ica Features

Abstract

Achieving performance and robustness simultaneously for an automated biometric system is crucial to prove that it is robust to real time noise encounters which determines the expected behavior of the system with the employed feature extraction algorithms, feature selection rules, classifiers. We have evaluated our system on two hand based modalities such as palmprint and handvein. The robustness is checked and the systems security level is explored on both unimodal and multi-modal (pre and post classification fusion techniques) systems by tabulating the comparative results got from clean and noisy data (Gaussian, salt and pepper and speckle noise) under the standard benchmark threshold values 0.01%, 0.1%, 1%.