Face and Palmprint Biometric Recognition by using Weighted Score Fusion Technique

Abstract

A multimodal fusion biometric verification system for face and palmprint modalities is proposed. The goal is to achieve a higher Accuracy for standard Databases. Fusion is done at score level using feature extraction algorithms such as, Radon transform, Ridgelet transform, TPLBP, FPLBP HOG, Gabor filter and DCT. Experiments are conducted on face94, face95, face96, FRGC IITD and PolyU databases. Only 1 image is given as a training set for each subject in respective databases. Matching Algorithm is used so as to achieve maximum GAR (Genuine acceptance rate). The results are discussed further in the paper. The accuracy achieved is 99.6% for FAR (False Acceptance rate) of 0.1%. Experimental results indicate that this approach although simple yet can achieve a greater accuracy.