Investigating Effect of Score Level Fusion on Palmprint Palm-vein Multimodal Biometric Identification

Abstract

Identification using biometric systems has many advantages. Moreover, multimodal biometric systems comprising more than one biometric characteristic, improve the accuracy of identification. Palmprints and palm veins are among the physical biometrics with the most biometric features. Palmprints and palm veins can be used as biometrics since each person has unique and stable features. They are distinct for different persons and can be acquired in a touchless acquisition setup. In order to use these biometrics for identification, first respective images have to be acquired and then the images need to be subjected to image processing stages such as pre-processing, extracting ROI, feature extraction and feature matching. This study proposes a method for segmenting the Region of Interest from the middle region of the palm by precisely aligning the palm in a vertical orientation with fingers facing upwards. For investigating the performance of identification, texture and appearance based features are extracted from the ROIs to form a feature vector. Weighted Euclidean distance is used for feature matching. Then the similarity score using Min-Max normalization are determined. Fusion is implemented using weighted sum method. The effect of score level fusion on identification is studied and evaluated with biometric parameters GAR and EER. The results obtained are analyzed in this paper.