Biometric recognition by multimodal face and iris using FFT and SVD methods With Adaptive Score Normalization

The main goal of this paper is to combine several biometric modalities (face and iris of a person) to improve recognition performance. For each modality, for features extraction, two approaches are used: Fast Fourier Transform (FFT) and singular value decomposition (SVD). Then, Classification is employed by Euclidean distance measurement. The most widely used normalization method such as min-max and z-score and a new method prctile (Percentiles of a data set). The fusion is performed at the score level with four methods such as: simple sum, weighted sum, min and max. Databases of 40 people, extracted from Olivetti Research Laboratory face Database (ORL) and the China Institute of Automation iris database (CASIA), are used for learning and testing the proposed system. Test results show that FFT (face) and FFT (iris) fusion scenario associated with min rule and new method (Prctile) gives the largest recognition result of 98.33%.